W&M ScholarWorks

Dissertations, Theses, and Masters Projects Theses, Dissertations, & Master Projects

2000

Surface processing by RFI PECVD and RFI PSII

Lingling Wu College of William & Mary - Arts & Sciences

Follow this and additional works at: https://scholarworks.wm.edu/etd

Part of the Condensed Matter Physics Commons, Materials Science and Engineering Commons, and the Plasma and Beam Physics Commons

Recommended Citation Wu, Lingling, "Surface processing by RFI PECVD and RFI PSII" (2000). Dissertations, Theses, and Masters Projects. Paper 1539623997. https://dx.doi.org/doi:10.21220/s2-sazs-yg02

This Dissertation is brought to you for free and open access by the Theses, Dissertations, & Master Projects at W&M ScholarWorks. It has been accepted for inclusion in Dissertations, Theses, and Masters Projects by an authorized administrator of W&M ScholarWorks. For more information, please contact [email protected]. INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6* x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

Bell & Howell Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA 800-521-0600

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with with permission permission of the of copyright the copyright owner. owner.Further reproduction Further reproduction prohibited without prohibited permission. without permission. SURFACE PROCESSING BY RFIPECVD AND RFI PSD

A Dissertation Presented to

The Faculty of the Department of Applied Science The College of William and Mary in Virginia

In Partial Fulfillment Of the Requirements for the Degree of

Doctor of Philosophy

by Lingling Wu

October, 2000

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number. 9999060

Copyright 2001 by Wu, Lingling

All rights reserved.

UMI*

UMI Microform9999060 Copyright 2001 by Bell & Howell Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

Bell & Howell Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPROVAL SHEET

This dissertation is submitted in partital fulfillment of the requirements for the degree of

Doctor of Philosophy

^Cingling^Wu

Approved, October 2000

Dennis M. Manos, Ph.D., Chair Department of Applied Science Ri £ W — Roy L. Champion, Ph.T). Department of Physics

H. Frederick Dylla, Ph.D. Thomas Jefferson National Accelerator Facility

Mark K. Flinders, Ph.D. Department of Applied Science

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table of Contents

Acknowledgments ...... vii

List of Tables ...... x

List of Figures ...... xi

A bstract...... xxi

1 Introduction ...... 2

1.1 Introduction to Surface Processing by RFI PECVD and RFI P S II ...... 2

1.2 Organization of the Dissertation ...... 8

2 DLC and Ti/MoS 2 Films Deposition on Ti Alloy for Tribological Applications ...... 14

2.1 Introduction ...... 14

2.2 PECVD DLC on Ti-6A1-4V ...... 16

2.2.1 Introduction to DLC Films ...... 16

2.2.2 DLC Films Deposition by Thin Film Processing Techniques ...... 18

2.2.3 DLC Films Deposition by RFI PECVD ...... 19

2.3 PVD Ti/MoSo Films on Ti-6A1-4V ...... 21

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 63 63 54 54 58 61 50 51 60 47 25 25 26 27 24 22 24 24 22 21 ...... iv ...... 2 Films Deposition for Triboiogical ApplicationsTriboiogical for Deposition Films 2 S o

3.4.1 3.4.1 Measurement Probe Langmuir 3.4.2 Spectroscopy Optical Emission 3.3.4 3.3.4 Confinement Magnetic 3.3.1 3.3.1 Window Dielectric 3.3.2 Antenna RF 3.3.3 Mode Capacitive Versus Mode RF Inductive 2.5.2 2.5.2 T Hardness est 2.5.3 Test Pull 2.5.4 Study SEM 2.5.5 Tests Triboiogical 2.5.1 2.5.1 of DLC FilmsStudy Ellipsometry 2.3.2 PVD Sputtering Fibns Magnetron by Deposition Ti/MoSo 2.3.1 2.3.1 M . 1 1 . Introduction .2 .2 RF ICP Source 3.4 ofRF IC Characterization P 3.3 3.3 and Building Design ICP Large-scale RF 2.7 2.7 Conclusion and Summary 2.6 2.6 and Discussion Characterization Results 2.4 Coating Preparation Cu-Ni-In Plasma Sprayed 2.5 2.5 Characterization Ti/MoSoFilms DLC and ro rn 3 3 Design and Characterization ofLarge-Scale RFICP Source 50 Reproduced with permission ofthe copyright owner. Further reproduction prohibited without permission. 3.4.3 Consideration of Sheath S ize and Arcing for Plasma Source Immersion Ion Implantation ...... 78

4 MAGIC Modeling of PSII Process ...... 89

4.1 Introduction to PSII Modeling ...... 89

4.2 Modeling of PSII Process ...... 92

4.2.1 Collisionless Model of PSIIProcess ...... 92

4.2.2 Modeling of PSII Process ...... 95

4.3 Introduction to MAGIC Code ...... 95

4.4 MAGIC Modeling of PSII Process ...... 96

4.4.1 Potential Structure, Field Distribution, Ion and Electron Distribution ... 98

4.4.2 Ion Flux Distribution ...... 109

4.5 Summary...... 116

5 Pattern Writing by Implantation in a PSII System ...... 120

5.1 Present Direct Write Technology ...... 120

5.2 Pattern Writing by Implantation ...... 123

5.2.1 Strategy and Advantages ...... 123

5.2.2 Experimental Conditions ...... 124

5.2.3 Results and Discussion ...... 126

5.3 Conclusion ...... 135

6 Deposition of N-doped SiC >2 Films by Novel Control of the rf Operation Mode ...... 138

6.1 Introduction ...... 138

6.2 Experiments ...... 140

v

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.3 Results and Discussion ...... 143

6.3.1 Operation of the PSII/PVD System ...... 143

6.3.2 AES and SIMS Depth Profiling ...... 144

6.3.3 TEM Study of the Cross Section of Processed Si Gratings ...... 145

6.3.4 Ellipsometry and Surface Profilometry Study ...... 151

6.4 Summary and Conclusion ...... 155

7 Surface Processing by Plasma Source Immersion Ion Implantation ...... 158

7.1 Introduction ...... 158

7.2 Experiment ...... 162

7.3 Results and Discussion ...... 165

7.3.1 Target Temperature ...... 165

7.3.2 Profile Calculation of Nitrogen Concentration ...... 180

7.3.3 Characterization of Implanted Samples ...... 184

7.4 Summary ...... 216

8 Conclusions and Future Work ...... 219

8.1 Conclusions...... 219

8.2 Future W ork ...... 224

References ...... 228

V ita ...... 238

vi

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Acknowledgments

I would like to thank Dr. Dennis Manos, for years of guidance and encour­

agement. I would also like to thank scientists and engineers of the FEL program

at Thomas Jefferson National Accelerator Facility (TJNAF), especially Dr. Fred

Dylla, Mr. Jock Fugitt and Mr. Richard Nelson, for their kind help with my experi­

ments. Thanks to Dr. Charles Sinclair, Mr. Timothy Siggins, Mr. George Biallas of

TJNAF, for their collaboration of FEL injection gun implants.

I wish to thank my committee members for their reading and criticism of this

manuscript. They are Dr. Roy Champion, Dr. Fred Dylla, Dr. Mark Hinders, and

Dr. Dennis Manos.

I would also like to express my appreciation to those individuals who wrote

the many letters of recommendation for me during my job search, including Dr.

Fred Dylla, and Dr. Brian Holloway. I am also grateful to Dr. Michael Kelley, who

gave me valuable advice on job search and career.

I want to thank Dr. Larry Ludeking of Mission Research Corporation, for

providing the code and training in the use of MAGIC.

Thanks to Mr. Richard Moore, formerly of Evans East, now of SUNY Albany,

for careful assistance with some of the surface analysis. 1 am also grateful to Dr.

David Cole of Evans East, for his help with SIMS analysis and with the data files.

vii

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thanks to Dr. Hongjun Gao, formerly of Oak Ridge National Laboratory, now of

Beijing Laboratory of Vacuum Physics, for performing TEM for us.

I want to thank Richard Proper, for helping setting up the lab and for help

with the experiments. I also want to thank Jay Diggs and Amy Wilkerson. for their

help, and for the time we spent together talking about wonderful things. Thanks

to Dr. Xianmin Tang for the help with OES study. Thanks to Carol Kalil for her

help with the AFM. I want to thank Dr. Edward Gillman of TJNAF, who has been a

wonderful friend and who has helped me with XPS and AES experiments. I want to

thank Casy Apeldoom and Dave McCay of the TJNAF machine shop, who quickly

turned so many of my designs into beautiful apparatus. I also want to thank Philip

Adderley and Ralph Afanador of TJNAF, for cleaning large vacuum parts and for

their help with vacuum systems. Thanks to Brandt Robertson of Old Dominion

University, for performing SEM and surface profilometry measurements for me.

I also want to thank Dr. George Loutts of Norfolk State University, for giving me

unlimited access to the XRD apparatus. I want to thank Dr. James Temer at Virginia

Commonwealth University, for his kind help with Raman measurements.

I want to thank Karen Berquist, for her help with my writing and for being a

great friend. Thanks to Cindy Manos, for helping me when I hurl my back and was

home alone. Thanks to Kathee, Bernadette, Paula, Dianne, and Sylvia, for taking

good care of me. Thanks also to the many graduate students, with whom I have

spent the last six years of my life. Thanks to Ingrid and Chris, for being such good

friends for so many years.

viii

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Finally, I want to express my most sincere appreciation to my family. Thanks

to my dear grandmother, who has been watching over me in the heavens. Thanks to

my mother, father, and brother, for believing in me and supporting my endeavors. I

would especially thank my husband, Xiangyu Huang, who has witnessed me getting

each and every degree, for years of understanding, help, and encouragement. Their

love and support mean a great deal to me and I could not have done this without

them.

ix

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. List of Tables

Table 2.1 Pull test results obtained using Sebastian I V ...... 36

Table 2.2 Hardness test results obtained using a Type M Shimadzu Micro hardness Tester...... 40

Table 3.1 The vibrational bands identified for the first positive system of excited nitrogen molecule [R. W. B. Pearse and A. G. Gaydon, The Identification of Molecular Spectra, Chapman and Hall, 1976] ...... 75

Table 3.2 The initial and final states of the vibrational bands identified for the second positive system of excited nitrogen molecule [R. W. B. Pearse and A. G. Gaydon, The Identification of Molecular Spectra, Chapman and Hall, 1976] ...... 78

Table 5.1 Nitrogen implanted layer thickness on Si substrate obtained by ellipsometry measurements ...... 129

x

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. List of Figures

Figure l.l Free Electron Laser electron injection gun at Thomas Jefferson National Accelerator Facility...... 4

Figure 2.1 Planar coil rf inductively coupled plasma enhanced chemical vapor deposition (rf ICPECVD) system used to deposit DLC films reported in this dissertation. The rf frequency used was 13.56MHz ...... 20

Figure 2.2 Magnetron sputtering deposition system used to coat Ti-6A1-4V substrates with 75%Ti/25%MoS2 and 25%Ti/75%MoS2 films ...... 23

Figure 2.3 Ellipsometry study of PECVD DLC film on Ti-6A1-4V substrate using J. A. Woollam M44 variable angle spectroscopic ellipsometer with WVASE32r

Figure 2.4 Real and complex index of refraction versus wavelength extracted from Ellipsometry data modeling. The best-fit thickness is 4230±60 angstroms. Typical index of refraction of single crystal diamond is 2.42 at 5890 angstroms ...... 31

Figure 2.5 SEM picture of PECVD DLC coated Ti-6A1-4V substrate ...... 32

Figure 2.6 SEM picture of 25%Ti/75%MoS2 coated Ti-6A1-4V substrate ...... 33

Figure 2.7 SEM picture of plasma-sprayed Cu-Ni-In coated Ti-6A1-4V substrate. . 34

Figure 2.8 SEM study of PECVD DLC coated Ti-6A1-4V substrate surface after pull test...... 35

xi

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 2.9 SEM study of plasma-sprayed Cu-Ni-In coated Ti-6A1-4V substrate surface after pull test ...... 37

Figure 2.10 Mass loss data for the bare and coated TU6AL-4V blocks obtained from block-on-ring tests ...... 38

Figure 2.11 Wear loss data for the bare and coated blocks obtained from block-on-ring tests ...... 39

Figure 2.12 Coefficient of friction data for the bare, plasma-sprayed Cu-Ni-In, and magnetron sputtered 25%Ti/75%MoS2 and 75%Ti/25%MoSo coated Ti-6A1-4V substrates ...... 41

Figure 2.13 SEM study of wear scar on the bare Ti-6A1-4V substrate after block-on-ring test ...... 43

Figure 2.14 SEM study of PECVD DLC coated Ti-6A1-4V substrate after block-on-ring test ...... 44

Figure 2.15 SEM study of wear scar on 25%Ti/75%MoSi coated Ti-6A1-4V substrate...... 45

Figure 2.16 SEM study of wear scar on plasma-sprayed Cu-Ni-In coated Ti-6A1-4V substrate after block-on-ring test ...... 46

Figure 3.1 Hot filaments and DC glow discharge plasma source used for the CWM PSII...... 52

Figure 3.2 Inductively driven sources in cylindrical and planar geometries ...... 53

Figure 3.3 Coordinate systems for a plate ...... 56

Figure 3.4 Impedance matching circuits for inductively coupled plasma ...... 59

Figure 3.5 The Faraday shield made for the PSII system ...... 62

Figure 3.6 CWM RFI PSII system, showing the Langmuir probe for measurement of plasma density profiles ...... 64

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.7 Measured ion plasma density for different rf power and nitrogen gas pressure without magnetic confinement. The probe was positioned 23 cm below the dielectric window...... 66

Figure 3.8 Radial ion plasma density distribution for nitrogen plasma at 1200 watts rf power and 5 x 10-4 Torr without magnetic confinement. The probe was positioned 23 cm below the dielectric window...... 67

Figure 3.9 Radial ion plasma density distribution for nitrogen plasma at 1200 watts rf power and 5 x 10-4 Torr with magnetic confinement. The probe was positioned 23 cm below the dielectric window...... 68

Figure 3.10 Ion plasma density for different rf power and nitrogen gas pressure with the magnetic congfinement ...... 70

Figure 3.11 Ion plasma density obtained without magnetic confinement and Faraday shield for nitrogen plasma at dififemt pressure and rf power. The probe was positioned 23 cm below the quartz window...... 71

Figure 3.12 Optical emission spectrum taken from 500 nm to 1000 nm for nitrogen plasma of 2.5 x 10'3 Torr and rf power of 1200 watts. The prominent bands are identified to be belonging to the first positive system of neutral nitrogen molecules ...... 74

Figure 3.13 Optical emission spectrum for nitrogen plasma from 550 nm to 700 nm ...... 76

Figure 3.14 Second positive system for neutral nitrogen molecules detected from nitrogen plasma of 2.5 x 10“3 Torr and 1200 watts ...... 79

Figure 3.15 Identified bands for No first negative system ...... 80

Figure 3.16 Sharp atomic lines from N + detected from a 2.5 x 10-3 Torr and 1200 watts nitrogen plasma. The vibrational bands of nitrogen first positive system are also labeled ...... 81

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.17 Sheath position as a function of time for plasma ion density of 1 x 1010/cm3. The electron temperature is 4 eV and the applied pulse voltage was 100 KV...... 84

Figure 3.18 Sheath velocity calculated for nitrogen plasma ion density of 1 x 10l0/cm3. The applied pulse voltage is 100 KV...... 86

Figure 3.19 Sheath velocity calculated for nitrogen ion plasma density of 1 x 10 10/cm 3. The pulse voltage used is 100 KV...... 87

Figure 4.1 Planar geometry showing the formation of ion matrix sheath ...... 93

Figure 4.2 The plasma sheath distribution at T = 110.070 ns for a negative voltage starting at To = 2 ns. The long cylindrical target was sitting on the sample stage, at the center of the cylindrial chamber...... 99

Figure 4.3 Electrical field distribution at T = 100.062 ns ...... 101

Figure 4.4 Equipotential lines at T = 105.066 ns ...... 102

Figure 4.5 Plasma ion distribution at T = 105.066 ns, showing laminar structure near the side of the long cylinder and focusing near the edges ...... 103

Figure 4.6 Electric field distribution at T = 109.947 ns...... 105

Figure 4.7 Equipotential lines distribution at T = 109.947 ns ...... 106

Figure 4.8 Electron distribution at T = 109.947 ns ...... 107

Figure 4.9 Nitrogen ion distribution at T = 109.947 ns ...... 108

Figure 4.10 Sheath distribution at T = 109.947 ns ...... 110

Figure 4.11 Nitrogen ion distribution at T = 109.947 ns...... I ll

Figure 4.12 The definations of the radial location and depth for a hole for ion flux monitoring ...... 113

xiv

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4.13 Accumulated total ion flux distributions for the bottoms of the holes...... 114

Figure 4.14 The accumulated total ion flux distributions on the walls of three holes...... 115

Figure 4.15 Locations defined for monitoring accumulated total ion flux distribution for a trench ...... 117

Figure 4.16 The radial distributions of the accumulated total ion flux at the bottom of 2-inch deep holes ...... 118

Figure 4.17 Accumulated total ion flux distribution on the trench walls ...... 119

Figure 5.1 Implanted nitrogen concentration depth profile for Ti substrate obtained by AES depth profiling ...... 128

Figure 5.2 Implanted nitrogen concentration depth profile for Si substrate obtained by AES depth profiling ...... 130

Figure 5.3 Nitrogen concentration in Ti substrate calculated by high dose model of Profile Code. The implantation voltage is 50 KV and the dose is 1.0 x 1018/cm2 ...... 131

Figure 5.4 Nitrogen concentration in Ti substrate calculated by high dose model of Profile Code. The implantation volatge is 50 KV and the dose is 2.5 x 1018/cm2 ...... 132

Figure 5.5 TEM picture of cross-section of nitrogen implanted Si gratings, showing the nitrogen-implanted layer, the Si substrate, and the SiOi cap lay er ...... 134

Figure 5.6 MAGIC modeling of PSII process in a trench, showing positive nitrogen ion distribution at time 1.2 x 10~7s ...... 136

Figure 5.7 MAGIC simulation results of ion flux distribution along a trench surface. The trench is 2 in. deep, 3 in. wide at the bottom, with side wall joining the bottom at 53.1 degrees. The simulation time is 2.5 x 10-7 s ...... 137

xv

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 6.1 The integrated PSII/PVD system for large-area, uniform, doped SiOo films depostion ...... 141

Figure 6.2 AES depth profile of bare Si substrate ...... 146

Figure 6.3 AES depth profile of nitrogen-doped SiC >2 coating on Si substrate 147

Figure 6.4 AES depth profile of bare Ti substrate ...... 148

Figure 6 .5 A E S depth profile of nitrogen-doped S iO ocoated T i substrate...... 1 4 9

Figure 6.6 SIMS depth profile of nitrogen-doped SiC >2 coating on Ti substrate 150

Figure 6.7 TEM picture of cross-section of processed Si AFM gratings, showing the nitrogen-implanted layer, the Si substrate, and the SiC>2 cap layer...... 152

Figure 6.8 Ellipsometry study of the thickness and optical properties of N-doped Si ( > 2 film on Si wafer. Cauchy optical model was used to model the experimental data. The obtained film thickness is 2447±9 angstroms. The film is almost transparent, with index of refraction of 1.483 at 5086 angstroms. The value for fused quartz is 1.462 at the same wavelength ...... 153

Figure 6.9 Profilometry measurement of nitrogen-doped S iO o film thickness 154

Figure 6.10 FEL stainless steel test electrode processed by PSII/PVD. The electrode is 0.15 m ID ...... 156

Figure 7.1 Traditional ion implantation setup ...... 160

Figure 7.2 Concept of PSII ...... 160

Figure 7.3 CWM large-scale RFI PSII system ...... 163

Figure 7.4 Target temperature (in Kelvin) as a function of time in seconds. The applied pulse is 30 KV at 100 Hz. The target surface emissivity is 0.2 ...... 169

xvi

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 7.5 Target temperature (in Kelvin) as a function of time in seconds. The applied pulse is 30 KV at 100 Hz. The target surface emissivity is 1.0 ...... 171

Figure 7.6 Target equilibrium temperature for different pulse repetition rate and surface emissivity. The chamber temperature used is 400 K...... 172

Figure 7.7 Target equilibrium temperature for different pulse repetition rate and surface emissivity. The chamber temperature used is 300 K...... 173

Figure 7.8 Schematic of temperature history for the target. Time periods and temperature swings are exaggerated relative to typical PSII target temperature histories ...... 175

Figure 7.9 Target temperature history as a function of time. Assume that the target temperature has reached 600 K by pulse heating before cooling is turned on at t=0. Assume that 0.1 mm thick silver layer is used for heat conduction between the target and the cooled stage, which is at 298 K by active water cooling ...... 177

Figure 7.10 Target temperature history as a function of time. Assume that the target temperature has reached 600 K by pulse heating before cooling is turned on at t=0. Assume that 0.1 mm thick indium layer is used for heat conduction between the target and the cooled stage, which is at 298 K by active water cooling ...... 178

Figure 7.11 Calculated target temperature history showing the heating from a train of pulses. The target starting temperature was assumed to be 400 K. Eleven HV pulses were applied before the stage cooling was turned on (0 - 5.5 x 104us). The pulse repetition rate was 200 Hz. Cooling was turned on at 5.5 x 104us when the target was at 401.98 K...... 181

Figure 7.12 Target temperature history calculated showing the heating effect of individual HV pulses. The pulse repetition rate was 200 Hz...... 182

xvii

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 7.13 Target temperature history which considers the heating effect of individual HV pulse and the cooling from the stage. The pulse was 20 /is wide and the repetition rate was 200 Hz. The target temperature was 401.98 K after heated by 11 HV pulses. The cooling was turned on at T = 55000 /is 183

Figure 7.14 Profile Code calculation of nitrogen concentration in Ti, Nb, Mo and Ta substrates. The implantation voltage is 30 K.V and the implantation dose is 1 x 10 11 je rr?...... 185

Figure 7.15 Optical microscope picture of nitrogen implanted Ti substrate ...... 187

Figure 7.16 Optical microscope picture of un-implanted Ti substrate ...... 187

Figure 7.17 Optical microscope picture of un-implanted Ta substrate...... 188

Figure 7.18 Optical microscope picture of nitrogen implanted Ta substrate ...... 188

Figure 7.19 Optical microscope picture of nitrogen implanted Mo substrate...... 189

Figure 7.20 Optical microscope picture of un-implanted Mo substrate ...... 189

Figure 7.21 Optical microscope picture of un-implanted Nb substrate ...... 190

Figure 7.22 Optical microscope picture of nitrogen implanted Nb substrate ...... 190

Figure 7.23 Optical microscope picture of 0.03/im mirror finish 304 stainless steel substrate ...... 191

Figure 7.24 Optical microscope picture of nitrogen implanted mirror finish stainless steel substate...... 191

Figure 7.25 Surface hardness test results for nitrogen implanted and un-implanted Ti substrates ...... 193

Figure 7.26 Surface hardness test results for nitrogen implanted and un-implanted Ta substrates ...... 194

xviii

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 7.27 Surface hardness test results for nitrogen implanted and un-implanted Mo substrates ...... 195

Figure 7.28 Surface hardness test results for nitrogen implanted and un-implanted Nb substrate ...... 196

Figure 7.29 Surface hardness test results for nitrogen implanted and un-implanted mirror finish 304 stainless steel substrate ...... 197

Figure 7.30 AES spectrum of the implanted Ti surface ...... 199

Figure 7.31 Concentration depth profile of a nitrogen implanted Ti substrate obtained by AES depth profiling ...... 200

Figure 7.32 Profile Code calculation of nitrogen concentration in a Titanium substrate for implantation voltage of 30 KV and dose of 6 x 10 17/cm 2...... 202

Figure 7.33 Profile Code calculation of nitrogen concentration in a Titanium substrate for implantation voltage of 30 KV and dose of 1 x 1 0 i 8/ c 7t i 2 ...... 203

Figure 7.34 AES spectrum of a nitrogen implanted Ta substrate surface ...... 204

Figure 7.35 AES depth profile of a nitrogen implanted Ta substrate ...... 205

Figure 7.36 AES spectrum of a nitrogen implanted Mo substrate surface ...... 207

Figure 7.37 AES depth profile of a nitrogen implanted Mo substrate ...... 208

Figure 7.38 AES spectrum of a nitrogen implanted Nb substrate surface ...... 209

Figure 7.39 AES depth profile of a nitrogen implanted Nb substrate ...... 210

Figure 7.40 AES spectrum of a nitrogen implanted stainless steel substrate surface...... 212

Figure 7.41 AES depth profile of a nitrogen implanted stainless steel substrate. . . . 213

xix

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 7.42 Nitrogen implanted FEL test electrode sitting on a support flange. It is 2.5 in. ID. The highly polished stainless steel surface shows uniform color. Its surface is also free of arcing or any other kind of damage ...... 217

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Abstract

An RFI plasma enhanced chemical vapor deposition (PECVD) system and a

large-scale RF plasma source immersion ion implantation (PSII) system were de­

signed and built to study two forms of 3-D surface processing, PECVD and PSII.

Using the RFI PECVD system, Ti-6A1-4V substrates were coated with diamond­

like carbon films with excellent triboiogical and optical properties. As an inno­

vation, variable angle spectroscopic ellipsometry (VASE) was successfully applied

for non-destructive, 3-D, large-area triboiogical coatings quality investigation.

Based on the experience with the RFI PECVD system, a large-scale RFICP

source was designed and built for the PSII. Langmuir probe and optical emission

spectroscopy studies indicated that the RFI source produced stable, uniform, and

clean plasma. MAGIC code was for the first time used to model PSII process, ad­

dressing different target geometries and boundaries, materials, plasma parameters,

illustrated sheath formation and evolution, field distribution, ion and electron trajec­

tories, ion incident angles, and dose distributions, which are critical for PSII design

and understanding.

The RF PSII system was developed into a versatile large-area, uniform, 3-D

surface processing apparatus, capable of PSII, PVD, PECVD, and in situ surface

cleaning and interface properties modification, for multilayer, multi-step, and high

xxi

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. performance surface engineering. Using the R F I P S I I system, for the first time, PSII

was studied as a mask-based surface layer conversion technique, for pattern writing

by implantation as an alternative to current deposition-based and ink-based direct

write technologies. It operates at low substrate temperature, keeps the original sur­

face finish and dimensions, and avoids adhesion problem. A different operating

mode of the RF source was discovered to perform biased sputtering of high pu­

rity quartz, which turned theR F I P S II system into a novel integrated RF PSII/PVD

system for large-area, uniform, nitrogen-doped, and hydrogen-free S iO o films de­

position at low substrate temperatures. Nitrogen-doped S iO o films with excellent

optical properties were deposited on semiconductor, metal, and polymer substrates

with excellent adhesion. Ellipsometry was used again for non-destructive S iO o

coatings investigation. F E L test electrodes processed by P S I I /P V D showed sup­

pressed field emission. A group of transition metals and an F E L test electrode were

also implanted by nitrogen using the P S II mode and analyzed.

xxii

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SURFACE PROCESSING BY RFIPECVD AND RFIPSII

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 1

Introduction

1.1 Introduction to Surface Processing by RFI PECVD and RFI PSII

This thesis examines advances in two forms of surface processing. Surface processing has

been playing an increasingly important role over the past decades, because of the impor­

tance of surface properties for the total performance of the material. In many applications

and constructions, failure originates at the surface [1]. The applicability and commercial

value of a material can be greatly enhanced by surface treatments. General examples in­

clude anti-rellective and anti-scratch coatings for optical components, impact and wear

resistant coatings for cutting tools, bearings and gears, lubricating layers for moving me­

chanical parts, and protective layer for electronics devices. Special applications include

surface processing of high voltage stainless steel electrode to suppress field emission and

ensure reliable and consistent performance of electron guns, which is crucial for the opera­

tion of both the main accelerator ring and for the Free Electron Laser at Thomas Jefferson

National Accelerator Facility (TJNAF) [2]. Another example is the oxynitridation of Si

2

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. / Introduction 3

substrates by rapid thermal oxidation in N 2 O to produce a dielectric layer that significantly

retards the boron diffusion, to improve the microelectronic device reliability [3],

High voltage photoelectron guns are used at Thomas Jefferson National Accelerator

Facility for electron beam production. The Free Electron Laser at TJNAF is the brightest

FEL in the world, with an average power output of 1700 watts, compared to a record of

merely 10 watts before the FEL at TJNAF was built. A schematic of the FEL electron

injection gun is shown in Fig. 1.1. A highly polished hollow stainless steel structure about

8 in. in diameter is used as the cathode, which supports the molybdenum base holding a

GaAs wafer. A pulsed Laser is used to radiate the GaAs wafer for photoelectron generation.

The cathode is usually biased to high potentials of up to 500 KV, yielding a field gradient

of 3-6 MV/m. The photoelectrons are accelerated by the strong electric field and exit the

anode aperture to form an electron beam.

Higher operating voltage is highly desirable for enhancing the photoelectron gun

performance. Bunch charge and emittance are two important design parameters for an

electron gun. The surface charge per unit area on the GaAs wafer is directly proportional

to the electric field applied. For each Laser pulse, a small portion of the GaAs wafer is

radiated and a small number of photoelectrons is generated [2]. The charge per bunch

requirement for the FEL electron gun is 135 pC/bunch. High voltage is also necessary

to achieve low electron beam emittance, which is determined by the size of the electron

emitting area and by the solid angle subtended by the electron beam. The electron beam

will expand due to the space charge effects. However, higher accelerating voltage will make

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. / Introduction 4

Corona Shield

Ceramic Insulator

Vacuum Cathode Chamber

Anode Plate

Figure 1.1. Free Electron Laser electron injection gun at Thomas Jefferson National Accelerator Facility.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. / Introduction 5

the beam travel faster toward its final target, allowing less time for the beam to expand thus

yielding a lower emittance.

Presently, the operating voltage of the electron guns at TJNAF is limited by field

emission from the metal electrodes and the supporting structures. The same problem exists

for other field particle accelerators, electron sources, and DC high voltage structures. As

applied voltage increases, a small current begins to flow between electrodes, which usually

precedes the formation of arcs. Arcs usually cause surface damage that further enhances

field emission causing more arcs, resulting in lower operating field strength of the electron

gun. The electron emission and ultimate breakdown voltage between these vacuum insu­

lated metal electrodes is believed to be caused by reduction of local work function, or by

the enhancement of local field strength as a result of surface imperfections, such as sharp

points, inclusions, and foreign contaminants. Currently, extensive hand polishing is used

to enhance electrode performance, requiring a surface finish of 9 pm to 1 pm. The polish­

ing is very time consuming and difficult for these large 3-D structures. Even after extensive

polishing, the performance of the electrode is unpredictable, with turn-on voltages varying

by factors of 2 or more for similarly prepared electrodes. Previous study indicated that ni­

trogen ion implantation reduced field emission from 1 cm ID stainles steel test samples [2].

However, it is necessary that high voltage breakdown experiments be conducted on full-

scale electrodes in the TJNAF test facility, because of the so-called "Total Voltage Effect”

[2], and the non-linear relationship between breakdown field and gap voltage.

An important goal of this dissertation is to develop a large-scale, 3-D, uniform surface

processing technique for these high voltage electrodes to suppress field emission. It is also

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. / Introduction 6

important that the process keeps the original target dimensions, without thermal distortion

of the target or degradation of original surface finish.

Coating and ion implantation are two important types of surface processing tech­

niques. Physical vapor deposition (PVD), chemical vapor deposition (CVD), solution and

sol-gel techniques, and thermal spray processing are frequently used to produce coatings.

PVD uses vapor to transport a material from a solid or liquid source onto a substrate to

form coatings [4]. PVD methods provide mainly line-of-sight coverage and have difficul­

ties coating 3-D objects with complex shapes. Solution and sol-gel processes can produce

limited types of coating, although they are inexpensive and relatively fast to coat large or

complex substrates. The thermal-spraying process is physically aggressive, and it is diffi­

cult to fine tune the film structure and composition. On the other hand, CVD processes rely

on specific chemical reactions on or near the substrate surface to obtain desired film compo­

sition, thus allowing reproducible film composition and 3-D coverage of the target. Usually

a higher substrate temperature is beneficial to promote chemical reactions on the surface,

and to enhance the film deposition rate and the ability to coat complex shapes. Plasma en­

hanced chemical vapor deposition (PECVD) is widely used to produce high quality thin

films at higher deposition rates, such as DLC films [5-7] and silicon nitride films [8,9].

The ion implantation process is a low-temperature surface processing technique. Ions

are accelerated into substrate surface to change its atomic composition and structure, thus

changing the surface materials property. It causes minimal thermal distortion of the target,

no degradation of surface finish, and no dimensional change of the work piece. Unlike thin

film deposition processes, there is no bonding failure or surface layer delamination prob­

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I Introduction 7

lem. Implanted species concentration and profile are easily controllable. Ion implantation

is also a non-equilibrium process, producing new alloys not limited by classical thermody­

namics and diffusion kinetics.

Plasma source immersion ion implantation (PSII) technique was developed for high-

dose, three-dimensional, large-area surface processing, in order to overcome the line-of-

sight problem and the limitation of beam current of traditional beam line implanters [10-

12]. Instead of an ion beam, PSII uses a plasma of the species to be implanted as the ion

source. The target is immersed in the plasma. By applying high voltage negative pulses to

the target, positive ions in the plasma are accelerated and subsequently implanted into the

substrate surface. All the surfaces of the target exposed to the plasma are implanted simul­

taneously. There is no need for target tilting or rotation. PSII is especially advantageous

for large-area, uniform, three dimensional, and high-dose implantations.

PSII has been widely applied to materials processing. Different plasma sources have

been used, including hot filament [2,11,13], cylindrical coil RF inductively coupled plasma

[14], RF capacitively coupled plasma [11,15], microwave plasma [16], and glow discharge

[11]. To realize PSII technique’s potential of treating large work pieces, several large-scale

PSII systems have also been built, including systems at Los Alamos National laboratory

[15], Hughes Research Laboratories [13], and the University of Wisconsin [11]. However,

none of these large-scale PSII systems takes advantages of the planar coil RF inductively

coupled plasma (RFICP, here abbreviated to RFI) source, a source which is widely used

to generate low-pressure, high-density discharges and which demonstrates the potential for

large area processing and improved spatial uniformity [12,17]. Since RF power is usually

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.2 Organization of the Dissertation 8

coupled to the plasma inside the vacuum chamber through a quartz window, there are no

electrical connections or supporting apparatus inside the chamber to take up extra space or

complicate chamber geometry.

Rf inductively coupled plasma source has previously been cited in the literature for

its potential for large-area processing and improved spatial uniformity [12,17-20]. Induc­

tively coupled plasma deposition (ICP) also has the advantage of independent control of

the discharge power and substrate bias voltage, which is very important in controlling the

incoming ion energy to the substrate [12,17,20]. ICP systems are also easy to implement.

RFI PECVD and large-scale RFI PSII were designed, built, and used as the surface

processing tools for experiments presented in this dissertation. Both PECVD and PSII are

non-Iine-of-sight surface treatment techniques. With the large-area, high density, uniform

plasma generating capability of RFI source, large-area, uniform, 3-D surface processing

were possible using RFI PECVD and RFI PSII.

1.2 Organization of the Dissertation

This dissertation presents the results of surface processing by RFI PECVD, large-scale

RFI PSII, and large-scale RFI PSII/PVD. The RFI PECVD system is capable of large-area

uniform diamond like carbon (DLC) films deposition. For the RFI PSII and RFI PSII/PVD

work, a large-scale system was designed and built, capable of processing multiple small

samples and large 3-D targets, such as the 8 in. i.d. electrode used for FEL electron

injection gun.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.2 Organization of the Dissertation 9

To generate a large volume, uniform plasma in the system for 3-D processing, a

large-scale planar RFI source was designed and built, which is believed to be one of the

largest planar RFI sources ever built. The up-scaling of RFICP source was based on the

successful operation of the 10 in. i.d. RFICP source designed for the PECVD system.

The operation and characteristics of this RFI plasma source has been carefully studied by

optical emission spectroscopy (OES) and Langmuir probe. These efforts have been made

with the point of view that up-scaling should be included in process research to eliminate

difficulties that result from unexpected behaviors in upscaling the smaller system [1]. In

building the large-scale RFI, up-scaling caused special material problems, which required

the adjustment of the original design and called for other techniques to ensure successful

operation.

Previous study indicated that nitrogen ion implantation reduced field emission from

1 cm ID stainless steel test samples [2]. However, full-scale HV breakdown tests are nec­

essary to ensure successful operation under higher voltages for FEL electron injection gun.

The non-linear relationship between breakdown field and gap voltage, and the Total Voltage

Effect, requires such extreme testing [2]. The large-scale RFI source for the PSII system

makes it possible to perform 3-D uniform nitrogen ion implantation on actual FEL elec­

trodes, for full size HV breakdown tests at TJNAF.

During the course of this work, MAGIC modeling was used for the first time to

elucidate important PSII issues, such as sheath formation and evolution, field distribution,

ion trajectories across the sheath, and dose distribution. MAGIC is an electromagnetic

particle-in-cell (PIC), finite difference, time-domain code that solves Maxwell’s equations

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.2 Organization of the Dissertation 10

and the Lorentz force equations in 3-D space. This modeling was much more useful than

simple 1-D or pseudo 2-D analytical results, made under a variety of assumptions and

simplifications. It also permits direct comparison to alternative PIC modeling results of

other groups. MAGIC was found to be more convenient than working with legacy plasma

modeling codes or programming-restricted low-level PIC codes.

Pattern writing by implantation has also been studied as an alternative to current di­

rect write technologies, taking advantage of the large-scale RFI PSII. Direct write tech­

nology is under intensive study as an approach to integrate 3-D mesoscopic electronic

components in a conformal manner, at low-substrate temperatures, on a large variety of

substrates. By integrating advanced electronics, functional devices, and active components

into the substrate, one can lower the system weight and reduce cost [21]. Ultimately, ma­

chine and electronics can become one integrated system. Possible applications include

miniature passive components for micro-satellites, munitions, toys, and wireless commu­

nications such as cell phones and GPS, embedded sensors for condition-based maintenance

and for the medical industry. Large-scale PSII pattern writing is advantageous compared

to deposition-based and ink-based direct write. It can produce features in a true conformal

manner, at low temperature for a larger variety of substrates without adhesion problem or

surface finish degradation.

In the course of our work, we discovered another operating mode of the RFI PSII,

PSII/PVD. This mode yields a novel way to produce large-area, high quality Si02 films at

low-temperatures on a large variety of substrates, with good adhesion. Because of the si­

multaneous PSII, spatial nitrogen incorporation into Si02 films can be controlled, which

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.2 Organization of the Dissertation 11

is important for producing high quality oxynitrde multi-layers for microelectronics device

manufacturing [22]. Because the processing gas is nitrogen, the chance of hydrogen incor­

poration into the SiC >2 films is greatly reduced. Hydrogen-free, low-temperature-deposited

SiC>2 films are strongly recommended for insulating inter-layers in ULSIs and for gate in­

sulators in thin-film transistors [23]. It is also convenient to use PSII at different operating

voltages to perform in situ surface cleaning and interfacial property modification. A large

FEL test electrode has also been processed by the PSII/PVD mode. Preliminary test results

of the processed electrode in the TJNAF HV breakdown test facility showed much higher

applied field gradients (exceeding 24 MV/m) and much smaller dark current.

Chapter 2 reports high quality DLC films RFI PECVD on Ti substrates for tribologi-

cal applications. The deposition process did not require bond layer or surface pretreatment,

yet yielded DLC films with excellent adhesion, which improved the substrate wear prop­

erties. For comparison, Ti/MoS 2 films were also deposited by magnetron sputtering PVD

and tested along with DLC samples. As an innovation, ellipsometry was used to character­

ize tribo logical coatings. It is fast, convenient, and non-destructive compared to destructive

and time-consuming mechanical tests. It also provides the advantage of yielding large-area,

3-D surfaces studying of coating quality.

Chapter 3 describes the design and characterization of a large-scale RFICP source,

including dielectric window selection, rf antenna design and the effects of magnetic con­

finement. Langmuir probe studies of spatial plasma distribution and OES study of nitrogen

plasma species are reported and are discussed in detail..

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.2 Organization of the Dissertation 12

Chapter 4 presents MAGIC modeling of the PSII process. Modeling of selected

geometries to study field distribution, ion trajectories, sheath formation, dose distribution

are described in detail.

Chapter 5 reports pattern writing by implantation in the large-scale PSII system.

Semiconductor and metal substrates were implanted through masks to study pattern trans­

fer fidelity and lateral diffusion. Ellipsometry studies of implantation dose distribution is

reported. Si gratings with 2 fim trenches were also implanted and investigated by TEM to

study direct-write of small features. The results were also compared to predictions from

MAGIC modeling.

Chapter 6 reports high-quality, nitrogen-doped Si02 films deposition by PSII/PVD.

AES depth profiling was used to study the films on Si and Ti substrates. Optical properties

of the SiOo films were studied by ellipsometry. An FEL test electrode was also processed

by PSII/PVD to suppress field emission, yielding excellent preliminary results.

Chapter 7 describes high-dose nitrogen implantation of an FEL test electrode and of

a group of planar transition metal samples, including Ti, Ta, Mo, Nb and stainless steel

(SS). Depth profiles of concentrations, surface hardness studies, and target temperature

calculations are reported. Ti and SS substrates are given more detailed attention because of

their application as FEL HV electrodes.

Chapter 8 ends this dissertation with conclusions and suggestions for future work.

Large-scale RFI PSII and PSII/PVD offers unique capabilities and many possibilities for

large area surface processing. Recommendations and possible improvements are presented.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.2 Organization of the Dissertation 13

Each chapter also includes its own background literature reviews and comparisons to

previous studies.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 2

DLC and Ti/MoS2Films Deposition on Ti Alloy for Tribological Applications

2.1 Introduction

In the course of this thesis, we designed and constructed an RF ICPECVD apparatus which

we used to create tribological coatings on Ti-based alloy substrates. We also created similar

coatings by sputter deposition, plasma spray, and sol-gel process, which were compared to

the RFI-produced coatings. The goal of this portion of the thesis was to reduce friction

in naval engine bearings. The design and operation experience of the 10 in. i.d. RFI was

also found valuable as a prototype for designing the large-scale RFI for the PSII system,

which is presented in Chapter 3. To characterize the DLC coatings, as an innovation, the

use of ellipsometry was developed to perform non-destructive assessment of 3-D large-

area tribological coatings. This technique was found valuable for PSII implantation dose

distribution study and non-destructive evaluation of coatings on PSII/PVD processed large

3-D FEL test electrode, as described in Chapter 5 and Chapter 6.

14

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 DLC and Ti/MoS^Films Deposition on Ti Alloy for Tribological Applications 15

Titanium alloys are promising candidates for tribological applications because of

their low density, high specific strength, high stiffness and high modulus. Titanium al­

loys rank ahead of conventional bearing steels when evaluated by the ASTM figure of

merit (FOM) for materials in tribological applications. In addition to their resistance to

compressive failure and their strength per unit weight, Ti alloys are resistant to most com­

mon forms of corrosion. However, Ti alloys have rather poor surface properties in other

respects. While it is crucial for tribological materials to have low coefficients of friction

and long endurance lifetimes, Ti alloys have high coefficients of friction and low resis­

tance to wear. This is complicated by the fact that Ti alloys are not effectively lubricated

by conventional liquid, greases, and solid coatings [24].

Plasma-sprayed Cu-Ni-In coatings have been used to protect against fretting dam­

age on the pressure surfaces of Ti-6AI-4V alloy blade dovetails in gas turbine engines [25].

However, exposed surfaces showed moderate to severe signs of Cu-Ni-In delamination,

surface crumbling, and tribo-oxidation. It is believed that failure occurs because of defects

within the coating and surface embrittlement due to oxidation. In general, thermal-sprayed

coatings have high surface roughness, high porosity, low thermal stability, and low repro­

ducibility that may degrade their performance. These performance limitations demonstrate

the need for alternative coatings for turbine engines. Since physical vapor deposition (PVD)

and chemical vapor deposition (CVD) are capable of producing high-quality films with few

defects, they were used as the deposition methods for this study.

Alternative coatings for T1-6A1-4V should be dense, have good adhesion, oxidation

resistance and low coefficient of friction. Furthermore, the coatings should have a hard

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2 PECVD DLC on Ti-6Al-4V 16

primary phase to provide wear resistance and load-bearing capability and a lubricating

secondary phase to reduce the friction between the two contacting components if wear

debris forms [25,26].

Two kinds of films have been studied as alternative coatings for Ti-6A1-4V surface:

diamond like carbon (DLC) and Ti/MoSo. Diamond-like carbon (DLC) and MoS2 films

have different wear and lubrication mechanisms. Diamond is naturally very hard and its

friction is determined primarily by the elastic deformation of the contact region [26]. Thus

it has very low coefficient of friction. On the other hand, MoSo has lamellar crystal struc­

ture. It is made of layers each of which comprises a plane of Mo atoms arranged in hexag­

onal array situated between two hexagonal layers of S atoms [27]. The attraction between

layers is weak, so the interfacial shear strength is low, giving a low coefficient of friction

[26]. However, the bonds between Mo and S atoms are covalent and strong, providing

excellent load capacity.

The study of both films will be presented in the following sections.

2.2 PECVD DLC on Ti-6Al-4V

2.2.1 Introduction to DLC Films

Diamond is unique natural material with a combination of extreme properties. It is the hard­

est known material, having extremely low coefficient of friction (0.05) and low abrasion

and wear. The surface of diamond is extremely stable and unreactive. Diamond has high

thermal conductivity and heat capacity that are comparable to copper, yet low coefficient

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2 PECVD DLC on B-6AI-4V 17

of expansion. It is also transparent to visible and infrared radiations. Diamond is a wide-

band-gap semiconductor, with a high breakdown electric field (4 x 106 to 2 x 10 7Vcm~l),

which is comparable to the breakdown field of -10 7Vcm~l for Si O2 [28,29].

Property Description Lattice constant 3.56725A (Silicon is 5.4307A) Density 3.515fc<7m-3 Coefficient of expansion 8 x 10“7 (fused quartz is 2.5 x 10-7) Thermal conductivity 1.5 x 104Wm~lK~l (copper is 4.2 x lO^Wm-1/^-1) Heat capacity 6.2KJkg~lK~l (copper is 4.2 KJkg~lK~l) Thermal stability Reacts withO 2 at > 600 °C) Young’s Modulus 1.054 x 10 (drawn tungsten is 3.5 x 10n Afm~2) Knoop hardness 7000 (Si 0 2 is 820) Dielectric constant 5.7 (silicon is 11.9) Band gap 5.5eV (silicon is l.leV) Resistivity > 1018Dm, type Ila, 3.4 x 10-5flm, type Ubb (copper is 2 x 10-8fim)

(“all properties are specified at 25°C unless otherwise stated.6Boron concentration is

2 x 1027m_3)

DLC films have been under intense investigation because of their valuable properties

[30]. Although DLC films do not have all the qualities of single crystal diamond, they do

exhibit some characteristics comparable to natural diamond. This makes them promising

for applications such as wear resistance and impact resistance coatings for cutting tools and

biomedical implants [31-33], bearings and gears, lubrication layers for moving parts, an-

tireflective and protective coatings for optical components, protective layers for electronic

devices, and heat sinks for high-temperature applications [34,35].

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2 PECVD DLC on Ti-6Al-4 V 18

2.2.2 DLC Films Deposition by Thin Film Processing Techniques

Researchers have used chemical vapor deposition (CVD) [5,36,37], plasma-enhanced

chemical vapor deposition (PECVD) [5-7], ion beam assisted deposition (IBAD) [38-40],

magnetron sputtering [41,42] and other novel methods such as laser ablation [43] and

plasma source immersion ion implantation (PSII) [44] to deposit diamond-like carbon. Of

the methods above, PECVD seems the most promising for depositing large-area, high-

quality films at high deposition rates [12,18,19]

Radio-frequency inductively-coupled plasma-enhanced chemical-vapor-deposition (rf

ICPECVD) was used in this work to deposit DLC films on Ti-6A1-4V substrates. Rf

ICPECVD has previously been cited in the literature for its potential for large-area process­

ing and improved spatial uniformity [12,17-20]. Inductively coupled plasma deposition

(ICP) also has the advantage of independent control of the discharge power and substrate

bias voltage, which is very important in controlling the incoming ion energy to the substrate

[12,17,20]. ICP systems are also easy to implement.

DLC coatings on Ti-6A1-4V have been studied previously by other groups, who used

ion beam assisted deposition processes [38-40,45]. Ion implantation gives better adhesion

than other surface coating processes, since there is no abrupt interface to suffer debonding

[31]. In two of the previous studies, a multi-layer coating with a nitrogen difiusion layer,

using a thin Si film to serve as a bonding layer, was needed to improve the adhesion be­

tween the DLC film and the substrate [38,39]. In the other studies, SiC was needed as bond

layer between the DLC and the substrate to overcome delamination of directly deposited

films [40]. Compared to ion beam assisted deposition techniques, rf ICPECVD is advanta-

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2 PECVD DLC on Ti-6Al-4V 19

geous because it does not require surface pretreatment (see below), it has large area coating

capability, it is a relatively simple coating process, and it produces films with excellent ad­

hesion. Most important, this coating method does not require a complicating bond layer

step.

2.2.3 DLC Films Deposition by RFI PECVD

We designed and built a special apparatus to perform ICPECVD of DLC for this work.

The system is shown schematically in Fig. 2.1. The top of the chamber is a 1.27 x 10-2

m thick quartz window. The rf antenna, which sits inside a steel cylinder for radiation

shielding, rests on the window during deposition. Methane and hydrogen flow into the

chamber through a side port. The distance between the heater stage and the quartz window

can be adjusted. To protect the rubber O-ring seals on the top of the chamber when the

substrate is heated, a water-cooled copper ring was added to the exterior seal edge.

For this study, three different substrates were coated at two different temperatures.

The substrates used were Ti-6A1-4V blocks (0.00635 mx0.0l016 mx0.01575 m), Ti-6A1-

4V coupons (0.01875 mx0.01875 mx0.003 m), and Rene 80 blocks (0.00635 mx0.01016

mx0.01575 m). Our Rene 80 samples were nickel alloy containing 13.7-14.3% chromium,

4.8-5.2% titanium, 2.8-3.2% aluminum, 3.7-4.3% tungsten, 3.7-4.3% molybdenum, 9.0-

10.0% cobalt, sampled from supplies used for turbine engine blades.

The surface finish of all substrates was 1.6 rms. All were coated without fur­

ther mechanical polishing. To minimize variation, all samples were coated simultaneously,

although the window-to-substrate distance varied slightly from the nominal 1.0 x 10-2 m

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2 PECVD DLC on Ti-6Al-4V 20

RF Radiation Cooing RF Power Shielding Water Supply Vlatch Box RF Antenna Cooling Water Water-cooed Quartz

Plasma Vacuum

Sample Stage with

I I I Methane and To Cryopump Hydrogen Gases To Mechanical Pump

Figure 2.1. Planar coil rf inductively coupled plasma enhanced chemical vapor deposi­ tion (rf ICPECVD) system used to deposit DLC films reported in this dissertation. TTie rf frequency used was 13.56MHz.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3 PVDi /MToS2 Films on Ti-6Al-4V 21

because of the different sample thickness. The substrate temperature before turning on the

plasma was ~800°C for the high-temperature DLC coatings. An rf power o f460 watts and

a mixture of about 95%H2, 5% methane at a pressure of 2.7 x 10- l Pa were used. For the

low-temperature DLC coatings, the substrate was heated to a temperature of ~400°C before

turning on the plasma. The deposition times were approximately 8 hours in both cases.

2.3 PVD Ti/MoS2 Films on Ti-6AI-4V

2.3.1 iMoSo Films Deposition for Tribological Applications

MoS2 is very effective as a solid lubricant [26,27,46] and it has been studied as lubricat­

ing coating on Ti-6A1-4V [24,47,48]. However, in the previous studies, described briefly

below, ion implantation and bond layers were necessary to modify the Ti-6A1-4V substrate

prior to MoS2 sputtering coating in order to increase the adhesion strength of the MoS2

film, and to increase the endurance lifetime [24]. Multi-layer coatings and various kinds of

pretreatments complicate coating procedures, increase costs, and invite process variability.

Mechanical polishing to remove surface oxides and improve surface finish [47,48] of Ti al­

loy parts is time consuming. In order to overcome these difficulties, in this work Ti/MoS2

films were deposited on the Ti-6A1-4V substrates using a simple two-step magnetron sput­

tering process. While MoS2 provides the lubrication, the existence of Ti in the coating is

thought to improve the adhesion of the film to the substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.4 Plasma Sprayed Cn-Ni-ln Coating Preparation 22

2.3.2 Ti/MoSs Films Deposition by Magnetron Sputtering PVD

A magnetron sputter deposition system was used to deposit the Ti/MoS 2 films for the study.

A schematic of the system is shown in Fig. 2.2.

Argon plasma was used for all depositions. Ti/MoS2 films were deposited at a pres­

sure of 1.3 x 10~l Pa - 6.7 x 10_1Pa, at an argon flow rate of 0.5-1.0 Pa m3/s, and a mag­

netron power of 100 watts. Typical sputter target bias was 400 volts dc. Two high-purity,

pressed powder targets, comprised of 25%Ti/75%MoSo and 75%Ti/25%MoSo, were used.

The composition was varied to test the effect of Ti in the MoSo film.

In-situ surface cleaning to enhance adhesion was performed by running a DC glow

discharge in the deposition chamber. The substrate was biased at -500 volts dc in a 1.33 x

101 Pa Ar/No environment (5% nitrogen). A current density of 1.0 x 102 A/m2 for 25 minutes

was adequate for Ti-6A1-4V substrates. This step removes some of the native oxide and

possibly produces a thin TiNx layer on the surface prior to deposition of the actual film,

both of which improve adhesion. The pre-treatment process was required for good adhesion

of Ti/MoSo coatings

2.4 Plasma Sprayed Cu-Ni-In Coating Preparation

Plasma-sprayed Cu-Ni-In coatings were deposited by an outside vendor [49] on Ti-6A1-4V

substrates. The gases used in the plasma spray were nitrogen and hydrogen, with primary

and secondary pressure both at 3.4 x 10s Pa. The flow pressure was 6.9 x 105 Pa. The

spray distance was 6.35 x 10-2 m and the spray rate was 7.6 x 10-4 kg/second [25].

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.4 Plasma Sprayed Cu-Ni-In Coating Preparation 23

Sputter Magnetron Sputter Deposition Chamber Target Assembly Substrate

Magnetron DC Bias Supply Power Supply

Gate Valve Rough Pump

A r ^ Turbo Pump

Figure 2.2. Magnetron sputtering deposition system used to coat Ti-6A1-4V substrates with 75%Ti/25%MoS2 and 25%Ti/75%MoS2 films.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5 DLC and TL/MoS2Films Characterization 24

2.5 DLC and Ti/MoS2Films Characterization

2.5.1 Ellipsometry Study of DLC Films

In order to provide a rapid and practical assessment of film coating quality and to indicate

whether further testing was warranted, we investigated the DLC films with optical methods.

Ellipsometry provides a fast, convenient, and non-destructive quality assurance step prior

to undertaking the more time-consuming and destructive mechanical analyses.

An M44 Variable Angle Spectroscopic Ellipsometer (J. A. Woollam Co. [50]) with

WVASE32r,u software was used to study the thickness of the DLC films. The Cauchy

Optical Model [51] was used to fit the experimental data. The method provides an accurate

measurement of film thickness. Since the optical properties of diamond are well known,

the comparison of the PECVD DLC on Ti-6A1-4V with diamond is a good indication of

sp3 content of the film. To our knowledge, this is the first time Ellipsometry has been used

to study DLC film quality. We also correlated these measurements with interference color

and the visible appearance of the films in an effort to provide a quick technical indication

of thickness and smoothness. This sort of practical guide has been found to be of value to

industrial practitioners [52,53].

2.5.2 Hardness Test

Film hardness was tested using a Type M Shimadzu Micro Hardness tester. At least 10

indentation measurements were averaged for each sample. Tests were conducted for DLC

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5 DLC and Ti/MoSzFilms Characterization 25

films of different thickness and Ti/MoS2 films. If not noted otherwise, the load used was

2.5 x lO-2 kg.

2.5.3 Pull Test

Film adhesion tests were done with a commercial pull tester [54]. The test procedure was as

follows: a hollow cylinder 2.8 x 10_3m in diameter, filled with epoxy, was secured perpen­

dicular to the face of the sample, using a fixture provided by the manufacture. The sample

and stud fixture were heated to 150°C for one hour, then cooled to room temperature. The

fixture was removed and the sample placed in the apparatus that pulled downward with a

measured, increasing force while holding the sample stationary. The instrument records

the highest value of force applied before failure. Microscopic examination of the delam­

inated areas was used to determine the actual contact areas, which were used to calculate

the tensile pressure required to delaminate the film.

Since the pull test is destructive, only a small number of samples were tested.

2.5.4 SEM Study

Scanning electron microscopy (SEM) was used to study the surfaces of bare and coated

Ti-6A1-4V substrates and the sample surfaces after pull tests and block-on-ring tests. All

samples, with the exception of adhesion test specimens, were rinsed with ethanol prior to

analysis to remove any surface contamination. SEM was used to confirm that all reported

pull test samples failed at the film-substrate interface and not from epoxy failure. Pull

test data was adjusted by calculating the actual pull-off areas of the coatings from the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5 DLC and Ti/MoSoFilms Characterization 26

SEM micrographs. SEM was also used to study the wear scars on the bare, PECVD DLC,

magnetron sputtered 25%Ti/75%MoS2, and plasma-sprayed Cu-Ni-In coated Ti-6A1-4V

blocks after the block-on-ring test.

2.5.5 Tribological Tests

Pin-on-disk Test

The friction and wear properties of the bare, plasma-sprayed Cu-Ni-In, and mag­

netron sputtered 25%Ti/75%MoS2 and 75%Ti/25%MoSo coated Ti-6A1-4V coupons were

tested (a suitable DLC sample was not available) using the pin-on-disk test (ASTM G99

procedure [55]). The normal force applied to the WC pin was 1 N. The disk rotated at

7.89 x 10~2m/s. The tests were conducted at ambient temperatures without any lubrica­

tion. Plots of the coefficient of friction as a function of elapsed time were obtained for each

specimen.

The wear tracks on the samples after pin-on-disk test were analyzed using profilome-

try. The wear track on the bare substrate was found to be -6.5 x I0_6m deep. The surface of

the plasma-sprayed Cu-Ni-In was too rough to accurately measure the track depth, although

a wear track was evident on the material. In contrast, the profiles of the 25%Ti/75%MoS2

and 75%Ti/25%MoS2 coatings did not show any detectable scarring, demonstrating im­

proved wear properties compared to bare substrate or plasma-sprayed Cu-Ni-In coating.

The lack of scarring on the Ti/MoS2 and the rough surface of the Cu-Ni-In made it im­

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 27

possible to quantitatively calculate a volume loss as was done for the block-on-ring test

below.

Block-on-ring Test

The friction and wear properties of the bare Ti-6Al-4V substrate, plasma-sprayed Cu-

Ni-ln, PECVD DLC, and magnetron sputtered 25%Ti/75%MoS2 coatings were also evalu­

ated using a block-on-ring test system (ASTM G77 procedure [56]). The 75%Ti/25%MoSo

coated sample was not available for this test. The frictional force required to keep the block

in place was continuously measured during the test with a load cell. The substrate block di­

mensions were 0.01575 mx0.01016 mx0.00635 m. The 52100 steel ring (0.0625 m OD,

and 0.0588 m ID) with a surface roughness of 2.2 x 10-7 m was heat-treated to a hardness

of 62 C Rockwell and surface finish of 6-12 rms. The rotation speed of the ring was 109

rpm. A load of 8.72 N was applied to the surface of the steel ring through a line contact of

3 x 10-3m. All tests were conducted at ambient temperatures without lubrication. For ac­

curate sample comparisons, each sample was tested for the same period of time under the

same conditions. The wear loss was reported as the volume loss in cubic millimeters, cal­

culated from the width of the block scar after the test. Mass loss data were obtained from

the pre- and post-test weights of the blocks and rings.

2.6 Characterization Results and Discussion

Optical measurements of the PECVD DLC films on both H-6A1-4V and Rene 80 substrates

show that they are highly transparent. When viewed under fluorescent room lights, the films

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 28

show bright interference patterns that change with different viewing angles. Fig. 2.3 shows

one typical set of such measurements, yielding a value of n around 1.9 (typical index of

refraction of single crystal diamond is 2.42 at 5890 A ), and a small extinction coefficient

(typically k<0.10), as shown in Fig. 2.4.

Ellipsometry indicates that DLC deposition yields smooth surfaces. This is confirmed

by SEM studies showing that the DLC (Fig. 2.5) is much smoother than the magnetron

sputtered Ti/MoS 2 or the plasma sprayed Cu-Ni-In (Fig. 2.6 and Fig. 2.7).

Table 2.1 shows pull test results for PECVD DLC, Ti/MoS 2 , and Cu-Ni-In coated

Ti-6A1-4V samples. The values in the table were calculated using accurate pull-off areas

measured from SEM photos, such as those shown in Fig. 2.8 and Fig. 2.9. The tensile

failure force for PECVD DLC on Ti-6A1-4V coupon was found to be 1.1 x 102MPa, indi­

cating a good bond between the DLC film and the substrate. The type of bonding between

the DLC film and the substrate determines the adhesion. DLC films deposited on transi­

tion metals such as Ni, Fe, and Co usually result in the formation of interposed graphite

layers causing poor adhesion [57]. It is believed that the excellent adhesion of the directly

deposited DLC film on Ti-6A1-4V can be attributed to the high hydrogen concentration

(95%) in the rf ICP. Adding a large percentage of hydrogen to the hydrocarbon gas feed

previously has been found to be beneficial in PECVD of DLC on various kinds of sub­

strates by aiding nucleation and growth [34], and reducing surface oxides allowing carbon

to react with the metal [36,37]. It has also been found that carbide formation facilitates

diamond growth [33,34,58-60].

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 29

PECVD DLC thin film coated samples exhibited higher hardness than the bare Ti-

6A1-4V and Rene 80 substrates. Table 2.2 shows the hardness test results for PECVD DLC

and Ti/MoS 2 coated samples. The variability of the results can be attributed in part to the

surface conditions of the substrates; the relatively rough 1.6 /im rms as-received substrate

surfaces were not polished. Since the substrates are much softer than the DLC films, the

Vickers test method criteria for film thickness to indentation depth were not rigorously sat­

isfied for these thin layers. The general problem of measuring the hardness of ultra-thin,

very hard layers over soft substrate has been discussed in reference [61]. Our measure­

ments can show qualitatively that the coated samples are indeed harder (probably much

harder) than the bare samples. The data shows that films with larger ratios of thickness to

indentation depth yield larger hardness values, confirming this qualitative result.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 30

Generated and Experimental Data

Model Fit Exp E 82° $ 36

4000 5000 6000 7000 8000 W avelength in A

Figure 2.3. Ellipsometry study of PECVD DLC film on Ti-6A1-4V substrate using J. A. Woollam M44 variable angle spectroscopic ellipsometer with WVASE32r,w software. Cauchy Optical Model was used to fit the experimental data. Shown is a typical set of ellipsometry measurements (dashed) and model fitting results (solid) of PECVD DLC films on Ti-6A1-4V substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 31

cauchy Optical Constants 1.925 0.10

n .c 1.920 c S 1.915 0.069 (0

1.910 0.041 X -a - 1.905

1.900 0.00 5000 6000 70004000 8000 W avelength in A

Figure 2.4. Real and complex index of refraction versus wavelength extracted from El­ lipsometry data modeling. The best-fit thickness is 4230±60 angstroms. Typical index of refraction of single crystal diamond is 2.42 at 5890 angstroms.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion

1 1 mm

Figure 2.5. SEM picture of PECVD DLC coated Ti-6A1-4V substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 33

Figure 2.6. SEM picture of 25%Ti/75%MoS2 coated Ti-6AI-4V substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 34

Figure 2.7. SEM picture of plasma-sprayed Cu-Ni-In coated Ti-6A1-4V substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 35

Figure 2.8. SEM study of PECVD DLC coated Ti-6A1-4V substrate surface after pull test.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 36

Sample Name Adhesion Strength Plasma-sprayed Cu-Ni-In 2.7 x 101 MPa High-temperature PECVD DLC 1.1 x 10* MPa 75%Ti/25%MoS2 7.6 x 101 MPa 25%Ti/75%MoS2 7.7 x 10lMPa Ti-6Ai-4Vwith 30-minute run of 75%Ti/25%MoS2 coating 8.5 x 10lMPa Ti-6Al-4Vwith 2-hour run of 75%Ti/25%MoS2 coating 7.9 x 10lMPa

Table 2.1. Pull test results obtained using Sebastian IV

Mass loss data and wear loss data from block-on-ring tests for bare, magnetron sput­

tered 25%Ti/75%MoS2, PECVD DLC, and plasma sprayed Cu-Ni-In coated Ti-6A1-4V

samples are shown in Fig. 2.10 and Fig. 2.11. PECVD DLC coated block has less wear

loss and mass loss than the bare Ti-6A1-4V substrate.

Magnetron sputtered Ti/MoS 2 films, like the DLC films, also have much smoother

and more uniform surfaces compared to the rough, porous surface of the plasma-sprayed

Cu-Ni-In coated substrate, as seen in Fig. 2.6 and Fig. 2.7. The hardness of 25%Ti/75%MoS

and 75%Ti/25%MoS2 coated samples is higher than that of the bare substrate (Table 2.2).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 37

Figure 2.9. SEM study of plasma-sprayed Cu-Ni-In coated Ti-6A1-4V substrate surface after pull test.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 38

0.005

Plasma-sprayed Bare Ti-6AMV CVDDLC Magnetron Sputtered Cu-Ni-ln 25%T!/75%MoSj

Figure 2.10. Mass loss data for the bare and coated T1-6AL-4V blocks obtained from block-on-ring tests.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion

0.7

0.6

0.5 co E E, 0.4 Mn o 0.3 an 0.2

0.0 Bare Ti-6AWV Sputtered CVD DLC Plasma-sprayed 25%Ti/75%M oSj Cu-Ni-In

Figure 2.11. Wear loss data for the bare and coated blocks obtained from block-on- tests.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 40

Bare substrate Film name Area color & Hardness Value hardness (MPa) Thickness (A) (MPa) 3500 ± 4 0 0 Low-temperature 6250 ± 120 Red 6300 ± 1200 (Ti-6A1-4V coupon) DLC 5280 ± 120 Green 6800 ± 1600 2760 ± 20 Sage 5100 ± 9 0 0 3400 ± 4 0 0 Low-temperature 9750 ±60 Red 4800 ±800 (Ti-6Al-4V block) DLC 8550 ± 70 Green 4300 ±600 6400 ± 2000 Low-temperature 7690 ±310 Red 7600 ± 1900 (Rene 80 block) DLC 13740 ± 70 Green 8700 ± 1200 3500 ± 4 0 0 High-temperature 3160 ± 30 Green 6500 ± 1600 (Ti-6A1-4V coupon) DLC 4230 ± 60 Purple 7400 ± 2200 3400 ± 4 0 0 High-temperature 5360 ± 50 Yellow 4900 ± 700 (Ti-6A1-4V block) DLC 5720 ± 50 Purple red 6200 ±900 3500 ± 400 25%Ti/75%MoS2 4100 ± 100 Ti-6A1-4V coupon 3500 ± 4 0 0 75%Tt/25%MoS2 4900 ± 3 0 0 Ti-6A1-4V coupon 3500 ± 4 0 0 Cu-Ni-In 4000 ± 1600 Ti-6A1-4V coupon

Table 2.2. Hardness test results obtained using a Type M Shimadzu Micro hardness Tester

With added Ti, the Ti/MoSo films retained the good lubrication properties and load-

bearing ability of MoSo, with the advantage of improved adhesion. The pull test data in

Table 2.1 show that magnetron sputtered Ti/MoS 2 coatings have much better adhesion (x3

) than the plasma-sprayed Cu-Ni-In. The film adhesion was found to be nearly independent

of film thickness in these runs, yielding similar values over a factor of four in deposition

time, as seen in Table 2.1. Because of the persistent presence of surface absorbed oxy­

gen and the difficult de-convolution of the Ti/Ti02 overlap in our XPS studies, reliable

quantitative data on Ti concentration in the films could not be extracted. We believe that

the percentages of Ti in the films for our two choices of target are similar. This implies

selective sputtering of the target must have occurred.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 41

1.00

0.80

0.60 2. Bare Ti-6Al-4V Substrate C £ £ 0.40 o c £ Plasma Sprayed Cu-Ni-ln £ 5 0.20 7S*Tt/25U1oS2 0 75XTt/25*«oS2

25%Tf/75W1oS2 25VH/75XMOS2 i i i i i i i r i 0 2 4 6 8 10 Elapsed Time, minutes

Figure 2.12. Coefficient of friction data for the bare, plasma-sprayed Cu-Ni-In, and magnetron sputtered 25%Ti/75%MoSo and 75%Ti/25%MoS2 coated Ti-6A1-4V substrates.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 42

Pin-on-disk test results indicate that the Ti/MoSo coated samples have better wear

properties than both plasma-sprayed Cu-Ni-In coated and bare Ti-6AL-4V substrates. As

shown in Fig. 2.12, the plasma-sprayed Cu-Ni-In coating, in current use to improve fret­

ting and wear properties of Ti-6A1-4V for turbine engine applications, has frictional prop­

erties very similar to the bare substrate. On the other hand, both Ti/MoS 2 coatings have

lower coefficients of friction than the Cu-Ni-In coating, with the 25%Ti/75%MoSo coat­

ing having the lowest coefficient of all samples tested. There was no detectable scarring

on either the 25%Ti/75%MoSo or 75%Ti/25%MoS2 coated samples after pin-on-disk test

compared to the 6.5 x 10-6m wear track depth on the bare substrate by profilometry and

the comparable wear track on the plasma-sprayed Cu-Ni-In. Block-on-ring tests show that

25%Ti/75%MoS2 coated sample has lower wear loss and mass loss than the bare substrate

(Fig. 2.10 & Fig. 2.11). We found the apparent disagreement in predicting performance for

Cu-Ni-In between two standard test methods represented in Fig. 2.10 and Fig. 2.11 some­

what puzzling. The wear loss for this coating would indicate that it is superior, whereas the

total mass loss would indicate that it is inferior.

The SEM study of the details of the wear scars from block-on-ring tests, revealed that

wear loss is not a good criterion for comparing wear properties of Cu-Ni-In coated sample

with other samples (Fig. 2.13, Fig. 2.14, Fig. 2.15, and Fig. 2.16). The apparent low

wear loss of plasma sprayed Cu-Ni-In resulted from the low wear scar width. However, the

width of the wear track does not give a full indication of potential service life. The plasma-

sprayed Cu-Ni-In coating, shown in Fig. 2.16, did not show either parallel scratch lines

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 43

Figure 2.13. SEM study of wear scar on the bare Ti-6Al-4V substrate after block-on-ring test.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 44

Figure 2.14. SEM study of PECVD DLC coated Ti-6A1-4V substrate after block-on-ring test.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 45

1 .Mill

Figure 2.15. SEM study of wear scar on 25%Ti/75%MoS2 coated Ti-6A1-4V substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6 Characterization Results and Discussion 46

iM £

Figure 2.16. SEM study of wear scar on plasma-sprayed Cu-Ni-In coated Ti-6A1-4V substrate after block-on-ring test.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.7 Summary and Conclusion 47

or clear wear tracks as the other samples did. The wear track was much deeper than the

scars for all other samples and the damage extended beyond the nominal contact area in an

irregular pattern. It appears that the plasma-sprayed Cu-Ni-In coating, which is both softer

and less well-adhered than PECVD DLC and H/M 0 S2 coatings, crumbled and detached

under load. Since the coating is non-uniform, and contains numerous particulates, the

coating delaminates irregularly near the wear track and the grit may further act to abrade

the coating. Thus ASTM wear loss calculated from the wear scar width is not a good

measure of the wear properties of plasma-sprayed Cu-Ni-In coatings. Actual mass loss

corresponds well with the full SEM analysis. The high mass loss of the test ring is likely

to be the result of excessive wear caused by the abrasive coating debris.

2.7 Summary and Conclusion

Ti-6AL-4V substrates were coated with DLC films (without surface pretreatment) and with

Ti/MoSo films (with only a simple surface pre-cleaning). Both kinds of films exhibited

much better tribological properties than the plasma-sprayed Cu-Ni-In coating that is cur­

rently used to protect Ti-6A1-4V against fretting and wear in turbine engine applications.

Variable angle spectroscopic ellipsometry was used to measure optical properties and

thickness of the PECVD DLC films. The optical properties of DLC film helped to quickly

identify the film and to indicate whether further destructive and time-consuming tests were

warranted. These measurements were also correlated with interference color and visible

appearance of the films to provide a quick assessment of thickness and smoothness. El-

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.7 Summary and Conclusion 48

lipsometry results showed that the DLC coatings were usually several thousand angstroms

thick. It was also noted that the thickness values directly corresponded with hardness val­

ues, but not with adhesion.

DLC and magnetron sputtered 25%Ti/75%MoS2 showed less wear and mass losses

than the bare substrates after block-on-ring tests (ASTM G77 procedure [56]) in agree­

ment with SEM studies. In contrast, the wear scar of the plasma-sprayed Cu-Ni-In coating

showed irregular crumbling and detachment of the coating in and around the contact area,

indicating poor wear resistance. Pin-on-disk friction tests were performed on bare substrate

and 25%Ti/75%MoS2,75%Ti/25%MoS2, and plasma-sprayed Cu-Ni-In coated substrates.

These tests indicated that the Cu-Ni-In coated substrate and the bare substrate have very

similar friction properties. Both Ti/MoS 2 coatings have better lubrication properties than

the bare and Cu-Ni-In coated substrates, with the 25%Ti/75% M 0 S2 coating having the

lowest coefficient of friction of all the samples tested by pin-on-disk method (ASTM G99

procedure [55]).

Results of micro-indentation tests indicate that the DLC and Ti/MoS2 coated samples

have greater hardness than both the Cu-Ni-In coated and bare substrates. Pull tests showed

that plasma-sprayed Cu-Ni-In coating has very poor adhesion (2.7 x 101MPa), compared

to PECVD DLC (1.1 x 102 MPa). The value is 7.6 x 10lMPa for 75%Ti/25%MoS2 and

7.7 x ltTMPa for 25%Ti/75%MoS2. Increased adhesion between M 0 S2 films and the Ti-

6A1-4V substrates may increase the endurance lifetime of the coatings [24]. The Ti/MoS2

coatings have exhibited excellent adhesion and improved wear properties as tribological

coatings on Ti-6A1-4V substrates. Under investigation by SEM it was found that DLC and

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.7 Summary and Conclusion 49

Ti/MoS 2 films look very smooth and uniform. On the other hand, the Cu-Ni-In coatings

were very porous, rough and non-uniform.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 3

Design and Characterization of Large-Scale RFICP Source

3.1 Introduction

Prior operation of the PSII system at the College of William and Mary with hot fila­

ment and DC glow discharge plasmas resulted in ion density estimated to be in the low

109/cm3range [2], in agreement with values previously reported in other large implanta­

tion systems [13,15]. Such low values of ti* made it difficult to produce the desired pulse

shape and to control their temporal profile. The average current per pulse was lower than

that designed for the pulse forming network (PFN)- Though higher densities in the fila­

ment driven source were achieved by increasing the filament temperature and bias, this led

to shortened filament life and the possibility of contamination of work pieces. Similarly,

sputtered contaminants made it difficult to increase density in a dc glow by raising voltage.

The requirement of long mean-free-path for the implanted ions eliminated pressure control

of rii as an option. In addition, both filament and dc glow discharge plasmas demonstrated

large non-uniformity over the large radial and vertical dimensions of this chamber. In some

50

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2 RFICP Source 51

of the applications, such as implanting large (~30cm), shaped, highly polished objects to

alter electrical properties, such non-uniformity is unacceptable. Although some effort was

made to improve spatial uniformity, such as multiple hot filaments on the supporting plate

at the bottom of the chamber (Fig. 3.1), its effects were not satisfactory.

To overcome these difficulties, as part of this thesis, a large RF inductively coupled

plasma (RFICP or RFI) source was designed and built. With a 0.43 m in diameter RF

antenna sitting on a 0.57 m in diameter quartz window, stable RFI plasmas can be generated

in the chamber.

3.2 RF ICP Source

Early inductive discharges had cylindrical configurations with coils wound around evac­

uated tubes [12]. A ring-shaped plasma resulted because of the limited skin depth of the

exciting field [17]. There was also controversy about whether these coils were capacitively

driven or inductively driven. Further studies recognized that the discharge was capacitively

driven at low plasma densities and switched to operate in an inductive mode at higher den­

sities. The planar coil configuration was developed in the late 1980s with the intent of

producing high-density low-pressure plasma for materials processing (p < 50 mTorr) in

low aspect ratio geometries ( l/R < 1 for a cylindrical chamber). Inductive discharges for

materials processing are referred to as TCP (transformer-coupled plasma), ICP (inductively

coupled plasma), or RFI (RF inductively plasma) [12].

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2 RFICP Source 52

DC Power Supply Hot Filament Power Supply

HV Feed Feedthrough f and Shielding ) Gas

Mechanical Turbo Punp Punp

Figure 3.1. Hot filaments and DC glow discharge plasma source used for the CWM PSII.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2 RFICP Source

c RF Power Input o Dielectric c

|. -ir-n. w V jCjt.-.yjrH

RF Bias

Planar Coil RF Antenna Dielectric • • •

Multipoles

Substrate "-T

Substrate ^ ------

Figure 3.2. Inductively driven sources in cylindrical and planar geometries.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building 54

The schematics of inductively coupled plasma sources with cylindrical and planar

geometries [12] are shown in Fig. 3.2. Both the helical and planar coils act as inductive

elements in the circuit, which typically consists of additional tunable capacitors to match

the coils to a 50f2 rf power supply. The dots in the planar geometry represent the cross-

sections of the planar rf antenna. Because of the plasma non-uniformity caused by the skin

depth of rf radiation, the cylindrical coil geometry produces strong radial variation in the

chamber, which is not good for uniform materials processing [12,18]. On the other hand,

the planar geometry has been found to be able to generate uniform plasmas over large

planar areas [12,17]. For a planar geometry, the planar spiral coil is separated from the

vacuum chamber by a dielectric window, which provides rf energy coupling and vacuum

seal. Magnetic multipoles are often added to confine the plasma, improve the plasma radial

uniformity, and raise the ion plasma density. ICP also allows for independent control of

the ion/radical flux and ion bombarding energy by biasing the substrate with a second rf

power supply. For most of studies, simple spiral coils have been used to generate uniform

plasmas [18]. More elaborate shapes have also been studied to change radial distribution

of the peak electron density [19],

3.3 Large-scale RF ICP Design and Building

3.3.1 Dielectric Window

A large rf coil was chosen to achieve large-area uniform plasma generation. Thus it is

necessary to use large dielectric window for rf coupling. For our 0.61 m ID chamber, it

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building 55

was possible to open a 0.44 m in diameter hole on the top flange, leaving enough space on

the flange to cut grooves for a double set of O-rings. It was decided that the appropriate

window size was 0.57 m in diameter, so that the quartz window can sit upon the two 0-

rings to make the vacuum seal with a pumpable interspace between the O-rings. Such

an arrangement greatly reduces the permeation of air and water and thereby improves the

cleanliness of the process. The quartz plate is simply supported along its contour by the

O-rings, pressing down by atmosphere pressure when the system is pumped, without any

clamping or extra weight.

It is important to calculate the safe thickness for such a large quartz window to be

strong enough to stand the pressure difference without breaking. Fused quartz has the

following typical mechanical parameters [29]:

Tensile strength 7,000 psi Poisson’s Ratio 0.16 Compressive Strength > 160,000 psi Rigidity Modulus 4.5 x 10 6 psi Young’s Modulus 10.4 x 10e psi Bulk Modulus 5.3 x 10e psi To estimate the stress on a 0.57 m plate exerted by I atm of air pressure, the following

calculations were made [62].

Fig. 3.3 shows the coordinate systems for a plate. The useful results for our consid­

erations are as follows:

The load and the supporting reactions are assumed to be normal to the middle plane

of the plate. The bending moments in the plate per unit length of the cross section can be

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building

Figure 3.3. Coordinate systems for a plate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building 57

defined as: f+ i Mi = / ^ X xzdz (Eq. 3.1)

r+i M,2 = / Yyzdz (Eq. 3.2) J- i X x and Yy are normal stresses.

Using a cylindrical coordinate system for a round plate that is simply supported along

its contour, the displacement of any point on the middle plane w can be calculated by :

and

Mi = -^[a2(3 + a) - r ( 3 + er)] (Eq. 3.4) lo

M* = ^ [ a 2(3 + a) - r 2(l + 3a)] (Eq. 3.5)

q =pressure, a =radius of plate, r =the radius of point of interest, a is the Poisson’s ratio,

which is 0.16 for fused quartz.

For a 0.57 m dielectric plate, considering the center of the plate, where r = 0,

1 013 x 105 Mi = - — ^ ------[0.2852 x (3 + 0.16) - 0 x (3 + 0.16)] = 1.625 x 103 N. (Eq. 3.6)

Where air pressure q = 1.013 x 10s Pascal, and

The maximum of a„ is found at the center of the plate:

3 CL 17max = g(3 + (r)q(-) (Eq. 3.7)

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building 58

Considering the tensile strength of quartz, which is 4.827 x 107 Pascal (7000psz),

hmin = 1-42 cm (0.560 in.).

It was concluded that for the diameter of our dielectric window, it is better to choose

0.0254 m (1 inch) thick quartz to have enough safety margin.

3.3.2 RF Antenna

Usually a capacitive network is required to match an rf inductive coil, to produce a resistive

50fi load to the rf power supply. The typical setup, shown in Fig. 3.4, consists of two

capacitors, one in series and one in parallel with the rf coil. The plasma usually acts as a

resistive load in the matching circuit

The admittance, looking to the right, at the terminals A - A! is [12]

YA=GA + jBA = Ra+j(Xi + Xs) (Eq- 3.8)

WhereG,\ = ^ i ^ x ’+x ^ ^ &A = ~ «*+(£+jr.)*•

At matched conditions,

Ga = ^ - = 1/5011 (Eq. 3.9) tvr

B a +WC2 — 0 (Eq. 3.10)

so that there is maximum power transfer from the power supply and the matching circuit

behaves as a 50Q resistive load to the power supply.

It is necessary to consider the tuning range of the available capacitors, the inductance

of the rf antenna, and the load behavior of the plasma, so that match can be achieved.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building

M atching RF Im pedance S o u rce

ICP Applicator Plasma -----

VNAA y w w .

+ *r L. Vt C,

A’

Figure 3.4. Impedance matching circuits for inductively coupled plasma

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building 60

The inductance of a spiral coil can be estimated from

B'W

L - i S T » E * ( E q ' 3 ' H )

where B =radius (inches), N =number of turns, C =thickness of spiral coil (inches)

Using this equation, it was calculated that for a 3 turn coil 8 inches in diameter,

L £ 8.63 fiH.

A three-turn rf spiral coil was made for our RFI source using quarter-inch copper

tube. The coil is cooled by water when operating. The two capacitors used are Jennings

vacuum tunable capacitors, each with an adjustable range of 20 pF - 320 pF.

3.3.3 RF Inductive Mode Versus Capacitive Mode

To achieve ion distributions that are closer to monoenergetic, plasma source immersed ion

implantation is usually operated in the sub-mTorr regime, where the plasma ion density is

substantially lower than for plasmas operated at higher gas pressures.

It has been found that an RFICP discharge at low pressures has a capacitively coupled

component, causing sputtering damage to the dielectric window and contamination to the

processing target [12,63]. Sugai et al. studied the electrostatic or capacitive coupling

of an rf antenna to inductive plasmas [64]. They found that capacitive coupling plays

an important role not only in the discharge mechanism but also in impurity release from

materials around the antenna. Since the inductive coil is driven in an electrical resonance

condition, it is expected that high potentials exist on the coil. Such rf potentials will lead to

capacitive coupling to the discharge in the chamber, which is characterized by high negative

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building 61

voltages between the dielectric window and the grounded wall. Ions are accelerated across

the sheath and hit the dielectric window at high energies, causing sputtering of the window

[17], It has been observed that weak capacitively coupled discharges oflen co-exist with

inductively coupled plasma sources at low absorbed powers. Since high potentials exist on

the inductive couplers, it is doubtful that any ICP is entirely inductively coupled [17].

A Faraday shield is usually used to reduce the capacitive coupling [17,65,66]. Such

shields are often made of a metal sheet or tube with cut slits, which shorts the capac­

itive fields and allows primarily the inductive excitation of the discharge. The Faraday

shield used in our PSII system is shown in Fig. 3.5. The slits prevent secondary rf current

from flowing through the shielding metal sheet, otherwise the metal sheet would act as a

short-circuited secondary winding to the primary coil, preventing the inductive field from

reaching the discharge inside the chamber.

3.3.4 Magnetic Confinement

Magnetic multipoles were placed around our implantation chamber to improve spatial

plasma density distribution. Each row of magnets consists of a series of strong rare earth

magnets arranged in alternating N-S directions. Each of the magnets is I inch by I inch,

1/2 inch thick. The magnetic held perpendicular to the square surface is 6800 Gauss. The

alternating magnets provide a line cusp near the chamber wall, preventing plasma loss,

increasing the plasma density, and improving spatial plasma uniformity [12,67].

External magnetic fields have also been found to improve the stability of low pressure

operation of RF ICP [68]. Power transfer efficiency generally decreases with decreasing

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3 Large-scale RF ICP Design and Building 62

Figure 3.5. The Faraday shield made for the PSII system.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization o f RF ICP 63

operation pressure. The magnetic field increases the electron density by decreasing the

radial transport of electrons, which helps to improve power transfer efficiency. A more

stable discharge is possible due to increased plasma electrical conductivity.

3.4 Characterization of RF ICP

The ICP source is shown in Fig. 3.6. The Faraday shield is separated from direct contact

to the coil by a 0.0254 m thick 0.57 m in diameter quartz plate. The shield is grounded at

two (or more) places on the circumference. The quartz window for making the vacuum

seal is 0.0254 m thick and 0.57 m in diameter. Permanent magnets are placed around the

chamber wall near the rf coil end. Processing gases are fed through the side port. The

system is pumped by a turbo molecular pump and mechanical pump (see chapter 7 on

surface processing by large-scale PSII).

3.4.1 Langmuir Probe Measurement

A Langmuir probe was used to characterize the ion plasma density and the radial density

distribution in the cylindrical chamber before and after applying the permanent magnets

around the chamber. Limited by the configuration of available ports, the Langmuir probe

was mounted through a port in the middle of the chamber on a motion feedthrough about

23 cm below the quartz window. The probe can be moved to different radial positions in the

chamber. The probe was connected to a bipolar power supply (Kepco), which was remote-

controlled by a computer, to provide bias voltages within a desirable range having pre­

programmed increments. The bias voltage and current through the probe were measured by

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 64

Quartz Insulating Plate Power Supply Cooling Faraday Water Shield

Permanent Magnets Feed Has PC Biasing Control HV Feedthrough Langmuir Probe Automatic Data Choke Acquisition Shielding Biasing

Valve

_ Mechanical Turbo Punp Pump z:

Figure 3.6. CWM RFI PSII system, showing the Langmuir probe for measurement of plasma density profiles

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RFICP 65

two multimeters, interfaced to a computer by GPIB board. Each run of the data acquisition

program collects a complete I-V curve and automatically saves the data in a file.

For the plasmas run at sub-mTorr to several mTorr pressure, coilisionless sheath con­

dition can be assumed. Thus a cylindrical probe was constructed and the following equation

was used to calculate ion plasma density [69,70].

I = 2enaad{2e ^ p ~ V b (Eq. 3.12) m

where e is electron charge, n3 is the density of the electrons or ions, a is the radius of

the cylindrical probe, d is the length of the probe, m is the mass of the ions or electrons

collected. V b is the probe bias voltage and I is the ion or electron saturation current.

Fig. 3.7 shows the ion plasma density at the center of the chamber under different rf

power and pressure without magnetic confinement. Fig. 3.8 shows the radial plasma ion

density distribution at 1200 w rf power and 5 x 10-4 mTorr nitrogen gas pressure without

magnetic confinement. It is apparent that the radial ion density distribution is approxi­

mately parabolic, as expected from unhindered diffusion within cylindrical geometry.

The radial distribution of ion plasma density with magnetic confinement is shown

Fig. 3.9. The ion density for different rf power and nitrogen gas pressure with magnetic

confinement is shown in Fig. 3.10. Compared to Fig. 3.8, the radial distribution of ion

plasma density is more uniform throughout the chamber, which is very important for large

area uniform implantations. The multi-cusp magnetic field added to the drift region at the

perimeter of the chamber provides a confinement field, which improves the spatial uni­

formity and increased the total ion density in the implantation chamber, allowing faster

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF [CP 66

n uE a>O 10 - «

400 600 800 1000 1200 1400 1600 RF Power (watts) • 5e-4Torr ----- Linear Regression ■ 1.0e-3 Ton------Linear Regression A 1.5e-3 Torr ------Linear Regression ♦ 2.5e-3 Ton- Linear Regression

Figure 3.7. Measured ion plasma density for different rf power and nitrogen gas pres­ sure without magnetic confinement. The probe was positioned 23 cm below the dielectric window.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization o f RF ICP 67

2* C/3 0Jc 0.9 - Q | 0.8 ■ mcn - 0.7- o "SN 0 .6 - 75 ^ 0.5 * wO z

0.3 ^»0 -30 -20 -10 0 10 20 30 40 Distance from the Center (cm)

Figure 3.8. Radial ion plasma density distribution for nitrogen plasma at 1200 watts rf power and 5 x 10-4 Torr without magnetic confinement. The probe was positioned 23 cm below the dielectric window.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 68

c 0.9

-20 -10 0 10 20 Distance from the Center (cm)

Figure 3.9. Radial ion plasma density distribution for nitrogen plasma at 1200 watts rf power and 5 x 10-4 Torr with magnetic confinement. The probe was positioned 23 cm below the dielectric window.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 69

processing times for PSII. This result is also in agreement with previous studies [71], per­

formed in smaller chambers.

The peak values of the ion plasma densities with and without magnetic confinement

are similar for different rf powers and gas pressures. However, as shown in Fig. 3.9,

magnetic confinement increased plasma uniformity and conductivity, and thus improved rf

coupling efficiency [67,68].

It is also interesting to note that the rf inductively coupled plasma densities at the

same powers and gas pressures are very similar to that with some degree of capacitive

coupling (Fig. 3.11). It can be explained partly by the fact that with the addition of a

Faraday shield and a second insulating dielectric window, the distance from the inductive

coil to the plasma is more than doubled. This results from the need to use the Faraday shield

to ground the high voltage on the inductive coil, which makes it necessary to introduce a

second quartz plate to separate the metal shield from the inductive coil. As the distance

between the coil and the plasma is increased, the mutual inductance between the coil and

the plasma decreases. The increased distance thus requires a larger inductive current in

the coil to sustain the discharge in the chamber. Another factor that influences coupling

efficiency is gas pressure. Low electron-neutral collision frequency for low pressure plasma

causes decreased plasma conductivity [67]. To maintain the same rf power absorption at

low pressures, the coil current has to be higher to increase the inducted electrical field in

the plasma. Since more power is dumped on the inductive coil and the capacitors at higher

current, the power transfer efficiency is decreased.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization o f RF [CP 70

Nitrogen Ion Plasma Density

1 4

12

O 10

8 (A

4

2 0 200 400 600 600 1000 1200 1400 1600 RF Power (watts) • 5a»4 Torr

■ 1 0«-3 Torr

▲ 1 5a-3 Torr

♦ 2.5e-3 Torr ---- -

Figure 3.10. Ion plasma density for different rf power and nitrogen gas pressure with the magnetic congfinement.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization o f RF ICP 71

Ion density versus RF power and Nitrogen pressure

600 800 1000 1200 1400 1600 1800 RF Power (watts) 0 5e*3Torr lt*3T orr 1 5e*3Torr

Figure 3.11. Ion plasma density obtained without magnetic confinement and Faraday shield for nitrogen plasma at diffemt pressure and rf power. The probe was positioned 23 cm below the quartz window.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 72

With the Faraday shield, the capacitive coupling fraction of our RFICP source is

insignificant. This has been confirmed by surface analysis of implanted samples prepared

using very long implantation times (see Chapter 7 on PSII of metals and semiconductor

substrates). Previous studies also found that thick dielectric windows help to reduce the

capacitive coupling between the coil and the plasma [17].

To initiate the discharge in the present RFICP system with large coil to plasma dis­

tance, hot filament and DC glow discharges are used. These discharges provide electrons

and ions required. Typically, a filament voltage of 4V and current of 1.5 A, a bias voltage of

100V was enough to start the ICP discharge. Once the plasma is ignited, the hot filament

and DC bias are not necessary to sustain the discharge.

3.4.2 Optical Emission Spectroscopy

It is important to understand the composition of the plasma in order to control the incoming

ion energy and to determine proper dose for plasma source ion implantations. Atomic

and molecular spectroscopy, which helped the development of quantum mechanics and the

understanding of structure of matter, can provide such information [72,73]. By studying

spectra, and by measuring transition probabilities, we can chart the energy levels. This

information can be used to identify species in the system. Depending on the energy, or

wavelength of the radiation from the transitions, different experimental techniques are used,

such as XPS, AES, x-ray spectroscopy, optical emission spectroscopy (OES), Raman, and

FTIR.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 73

Optical emission spectroscopy usually provides information about molecular vibra­

tions, and the transitions between outer orbitals of atoms. It has been widely applied to the

study of plasmas as an non-intrusive evaluation method. Spatially resolved OES has been

used to study the different spatial distributions of various species in dc-magnetron sputter­

ing discharges [74]. OES was also used to study the nitrogen species in the plasma used to

perform surface nitriding of metal targets, in order to adjust the operation parameters and

enhance the nitriding efficiency [75]. It was also used to study the influence of hydrogen

on the density of molecular ions in dc and dc-pulse glow discharges [76].

In our work, nitrogen plasma optical emission spectroscopy was performed using a

spectrometer equipped with 256 x 1024 CCD camera as a detector. Spectra were obtained

through a 6-inch view port located 25.3 cm below the dielectric window. Three different

gratings were used (300, 600, and 1200 grooves/mm) for survey scans and for detailed

studies. The entrance slit was set to 6 fim for all the measurements, yielding a resolution

full width at half maximum of 2 A. The wavelength calibration of the gratings and the

spectrograph were done using a mercury calibration lamp.

A typical spectra for the nitrogen plasma taken for 500 nm to 1000 nm range is shown

in Fig. 3.12. An enlarged portion for 550nm to 700nm range is shown in Fig. 3.13.

The prominent band structures appearing in this range belong to the first positive system

(£ 3rig - A3E+) of neutral nitrogen molecules [77]. Table 3.1 below shows the respective

initial and final vibrational states for the bands identified in Fig. 3.12 and Fig. 3.13.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 74

20000

891.2 8722

15000 -

738.7 750.4

727.3 7626

W 10000 ■ 775.3 716.5

5000 -

500600 700 800 900 1000 Wavelength (nm)

Figure 3.12. Optical emission spectrum taken from 500 nm to 1000 nm for nitrogen plasma of 2.5 x 10-3 Torr and rf power of 1200 watts. The prominent bands are identified to be belonging to the first positive system of neutral nitrogen molecules.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 75

A (nm) v ' v " A(nm) v ' v " 968.2 3 3 670.5 5 2 943.6 4 4 662.4 6 3 920.4 5 5 654.5 7 4 891.2 0 646.9 8 5 872.2 2 1 639.5 9 6 775.3 2 0 632.3 10 7 762.6 3 I 607.0 6 2 750.4 4 2 601.4 7 3 738.7 5 3 595.9 8 4 727.3 4 590.6 9 5 716.5 7 5 585.4 10 6 687.5 3 0 580.4 11 7 678.9 4 1 575.5 12 8

Table 3.1. The vibrational bands identified for the first positive system of excited nitrogen molecule [R. W. B. Pearse and A. G. Gaydon, The Identification of Molecular Spectra, Chapman and Hall, 1976]

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RFICP 76

5000 654.5 662.4

4000 .670.5

3000 678.9 639.5 6313 687.5 | 2000 595.9 c 580.4 601.4

1000 607.0

-1000 550 600 650 700 Wavelength (nm)

Figure 3.13. Optical emission spectrum for nitrogen plasma from 550 nm to 700 nm.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RFICP 77

The N2 second positive system (C3IIU - 2?3n g) [77] was also detected and is shown

in Fig. 3.14. The identified bands and the respective initial and final vibrational bands

are shown in Table 3.2. There are several strong lines that do not belong to the second

positive system. Detailed study around the 400 nm range reveals that these bands are from

the first negative system (£ 2£„ — X2E+) of N f, as shown in Fig. 3.15. Detailed study

was also carried out around 740 nm. Atomic lines of atomic nitrogen were detected. A

typical spectrum is shown in Fig. 3.16. The four strong, known N-atom lines identified are

at 746.8 nm, 744.2 nm, 742.3 nm, and 740.6nm.

OES results show that the dominant species in the large-scale RFI are excited nitrogen

molecules, singly ionized nitrogen molecules, and atomic nitrogen. Lines for singly ionized

nitrogen atoms were not detected in this study. A previous study used the N-atom 869 nm

line for analysis of a high density nitrogen plasma [78]. The 869 nm line was not usable

in this study due to strong presence of the overlapping band at 872.2 nm from the N2 first

positive system.

The degree of dissociation is directly dependent on the input power density. Blant and

coworkers have reported that in very high density RFI plasmas in nitrogen, the dominant

features belong to atomic nitrogen [78]. That study used a helical coil to surround a small

discharge tube, 25 mm ID having a volume of 45 cm3. The resulting rf power density in

that tube was 13.3 W cm-3. By contrast, the volume of our large PSII chamber is 0.149 m3.

With an rf power input of 1200 W, the rf power density is only 8.05 x 10~3 W / cm3, more

than three orders lower. In addition to the power density difference, helical and planar coil

rf discharge have different spatial distributions. The electron energy distribution function

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 78

A v ' v " A i / v " 420.1 2 6 367.2 3 5 405.9 0 3 357.7 0 1 399.8 1 4 353.7 1 2 394.3 2 5 350.1 2 3 389.5 3 6 337.1 0 0 380.5 0 2 315.9 I 0 375.5 1 3 313.6 2 1 371.1 2 4 311.7 3 2

Table 3.2. The initial and final states of the vibrational bands identified for the second positive system of excited nitrogen molecule [R. W. B. Pearse and A. G. Gaydon, The Identification of Molecular Spectra, Chapman and Hall, 1976]

is very different in a small diameter discharge tube, in which diffusion to the wall is an

important electron energy loss mechanism. In a large-volume chamber, such as the one

used for the large-scale PSII, electron-ion recombination is much more important [76].

It is important to note that no SiO, SiO+ or Si02 lines were detected, indicating that

any sputtered contaminants from the quartz window were below the sensitivity of the spec­

trometer used. This was also confirmed by AES results of the implanted samples (see

Chapter 7 on ion implantation by large scale RFI PSII), which showed no Si02 contamina­

tion.

3.4.3 Consideration of Sheath Size and Arcing for Plasma Source Immersion Ion Implantation

With the obtained experimental values of ion plasma density, the sheath size for plasma

source immersion ion implantation was calculated for experiment design. Assuming that

n, = 1 x 1010/ cm3, the ion matrix sheath width can be calculated for a planar geometry by

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 79

800 357.7 337.1

600 - 315.9 380.5 375.5

400 - 389.5 £ 313.6 353.7 371.1 u> 394 3 c Q) 367.2 C 311.7 200 - 350.1

-200 280 300 320 340 360 380 400 420 440 Wavelength (nm)

Figure 3.14. Second positive system for neutral nitrogen molecules detected from nitro gen plasma of 2.5 x 10-3 Torrand 1200 watts

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Character ization of RF [CP

400 391.4

300 -

423.6 388.4 427.8 >> 200 - ' c

-100 370 380 390 400 410 420 430 Wavelength (nm)

Figure 3.15. Identified bands for N f first negative system.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 81

600

746.1 500 -

744.2 400 - 742.3

300 - u 740.6 a>c c 200 -

100 -

-100 730 740 750 760 770 780 Wavelength (nm)

Figure 3.16. Sharp atomic lines from N + detected from a 2.5 x 10-3 Torr and 1200 watts nitrogen plasma. The vibrational bands of nitrogen first positive system are also labeled.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF [CP 82

solving Poisson’s equation, to yield:

s0 = (— )1/2 (Eq. 3.13) eng

It was found that s0 = 0.0182 m for Vo = 30 KV, s0 = 0.0235 m for Vo = 50 KV, and

So = 0.0333 m for Vo = 100 KV, all insignificant compared to the target and chamber size

used for our PSII system.

Understanding the temporal and spatial evolution of plasma sheath around real 3-

D targets during PSII requires computer simulation solving Maxwell’s equations and/or

Poisson’s equations (see Chapter 4 on MAGIC simulation). Sheridan et al. solved Poisson’s

equation numerically to calculate the expansion of the ion matrix sheath around a square

bar [79], A more comprehensive treatment has also been accomplished using a 2-D PIC

simulation [80], for time-dependent sheath formation, depletion of the initial ion matrix,

and for the expansion of the sheath into the ambient plasma. Such simulations are time-

consuming and complicated, so for simple estimates for experiment design, 1-D analytical

results are still widely used. The sheath width at the end of each pulse is an important

factor for uniform ion implantation. During implantation, ion current depletion can be

caused by overlapping of sheaths when multiple targets are installed, or by the sheath edge

reaching the chamber wall. The calculations below can provide important parameters when

large three-dimensional targets or multiple targets need to be implanted. The implantation

chamber must be larger than the sum of the 3-D target size plus twice the maximum sheath

width that develops within the applied pulse width. The distance between small samples

should be larger than twice the maximum sheath width. Note, however, that a large chamber

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 83

is not always desirable because it requires more power input to maintain a plasma. As the

ions in the sheath are implanted, the sheath recedes and more ions are recovered from the

moving sheath edge. It was found that the sheath velocity for a planar target geometry can

be calculated by [12,81]

I -Hr— ( & 1 ' 3 - l 4 )

where uo = ( ^ ) 1/2 is the characteristic ion velocity and ue = ( ^ f ) I/2 is the Bohm

(ion sound) speed. For the following calculations, Te was chosen to be 4 eV, since the

electron temperature in most discharges is clamped in the range of 2 - 5 eV by particle

balance [12].

Solving the above equation numerically for V0 = 100 KV, the sheath edge position

in meters versus time in seconds is shown in Fig. 3.17. The steady-state sheath width was

found to be 0.234 m. It is 0.139 m for 50 KV and 0.095 m for 30 KV.

From the plot, we see that the sheath width is 0.19 m at t = 20 /zs, which is the width

of the high voltage pulse used for the CWM PSII. Frequently, implantation voltages of 20

to 50 KV are used. Calculation shows that at t = 20 [is, s = 0.12 m for Vo = 50 KV,

s = 0.09 m for Vo = 30 KV. The implantation pulse width is usually adjusted to be shorter

than the time required for the sheath to reach the steady state Child law sheath thickness.

The sheath velocity was also calculated for 100 KV pulse and nitrogen ion plasma

density o fl x 1010/cm3. As shown in Fig. 3.18, the sheath velocity decreases dramatically

with time, falling to near zero velocity around 60 fis, when the steady sheath width is

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 84

0.2 -

0.15-

0.05

2e-05 0.0001

Figure 3.17. Sheath position as a function of time for plasma ion density of 1 x 1010/cm 3. The electron temperature is 4 eV and the applied pulse voltage was 100 KV.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 85

approached. Fig. 3.19 shows sheath velocity for the 20 ps to 200 ps range, calculated with

the same parameters as used for Fig. 3.18.

Collisionless sheath conditions require the mean free path to be longer than the sheath

width. Otherwise, the ions collide within the sheath, leading to reduced implantation en­

ergies and a broadened angular distribution of incoming ions, affecting the implantation

over surface topography. The mean free path, A, relates to the pressure by the expression

p x A = const, where the constant is 6.1 x 10-3 Pa • m for nitrogen [81]. This condition

gives a upper limit for the implantation pressure.

For a given implantation voltage, there is also an upper limit for the plasma density,

relating to the maximum electrical field in the ion matrix sheath by [12,81]:

m ax (Eq. 3.15)

The maximum electric field in the ion matrix sheath can not exceed a limit, which depends

on the materials, the working gas, and a number of other factors, otherwise, arcing will

occur. Arcing is highly undesirable for implantations on polished FEL electrode or on

sensitive substrates. This upper limit is approximately 10 KV/mm for gas pressure lower

than 1 Pa, although it also depends on the specific geometry and surface properties [81].

For the ion implantation process, this maximum field strength is reached at the beginning

of each pulse, when the sheath is narrowest. Using the above equation, for a 100 KV pulse

and for an ion plasma density of 1016/m 3, we calculate Em;iX = 6.01 KV/mm, lower than

the limit for arcing to occur. However, higher ion plasma density should not be used at such

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 86

160000- 140000-

120000 ■

100000- y 00000 60000:

2e-05 4e-05 *6e-05 8e-05 0.0001

Figure 3.18. Sheath velocity calculated for nitrogen plasma ion density of 1 x 1010/cm3. The applied pulse voltage is 100 KV.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 87

2000H

500-

4e-05 Be-05 0.00012 0.00016 0.0002 t

Figure 3.19. Sheath velocity calculated for nitrogen ion plasma density of 1 x 1010/cm3. The pulse voltage used is 100 KV.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4 Characterization of RF ICP 88

high implantation voltages if it is absolutely necessary to avoid arcing at the target surface,

thus we perform implants of polished targets at values lower than 100 KV in general.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 4

MAGIC Modeling of PSII Process

4.1 Introduction to PSII Modeling

Plasma source immersion ion implantation uses a plasma of the species to be implanted as

the ion source. Instead of continuous acceleration voltage, a train of high voltage negative

pulses, typically 1 - 100 KeV, 1 - 50/m long, are applied to the target immersed in the

plasma. The positive ions in the plasma are accelerated toward the target surface during

each pulse. The implantation dose distribution has been found to be strongly dependent of

plasma sheath distribution, plasma ion density distribution, and local electromagnetic field

distribution.

Shamim, et al., have studied the distribution of incident ions and retained dose for a

wedge-shaped target in plasma source immersed ion implantation [82]. It was found that

there was enhanced sputtering, less retained dose, and a shallower penetration range of

implanted ions in the vicinity of the acute angle edges as compared to square angle edges.

They also found that a smaller wedge face retained a higher average dose than a face with

larger area. The dose difference was thought possibly to be due to the stopping of the sheath

89

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 MAGIC Modeling of PSII Process 90

at different positions and to the difference in ionization rates of neutrals inside the sheath

due to different electric field intensities. Spatial and temporal sheath evolution of plasma

sheath for spherical and cylindrical geometries has also been measured and compared to

the calculations using the quasistatic Child-Langmuir approximation [12,83].

Scheuer, et al., also developed analytical models for planar, cylindrical, and spherical

geometries, assuming that the transient sheath obeys the Child-Langmuir law for space-

charge-limited emission at each instant during the propagation of the sheath, predicting the

sheath-edge position as a function of time, the final sheath extent, and average ion current

to the target during each pulse [84]. In contrast to the model proposed by Lieberman for

planar geometry [12], they assumed that an ion does not drift toward the target at the ion

acoustic speed prior to its propagating sheath, but rather has zero velocity until the sheath

edge reaches it. Scheuer argues that this is a more realistic model in the case of a fast

sheath since the presheath dose not have time to form before the sheath reaches a given

point. Later work used a one dimensional analytic model of dynamic sheath expansion to

develop a pseudo 2-D model [85].

Sheridan, et al., also simulated the ion matrix sheath surrounding a square bar and

the determined equipotentiai lines by solving Poisson’s equations, providing 2-D sheath

distribution for PSII [79]. Another study also predicted equipotentiai evolution and dose

distribution by solving Poisson’s equations for a cylindrical geometry [86], using finite dif­

ference method. Wang, et al., have developed a Monte Carlo simulation model to determine

the energy and angle distributions of ions incident on a spherical target during the sheath

edge expansion in higher pressure regime for PSII [87]. The group at Los Alamos [80] also

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 MAGIC Modeling of PSII Process 91

used PIC codes to study the sheath distribution and implantation dose distribution for long

rod and bore geometries. Paulus, et al., also used a PIC code to solve Poisson’s equations

around two simple shapes for ion current density, average impact energy, and other related

information [88].

These previous modelers were successful in providing information for some aspects

of PSII, such as implantation dose for each pulse for simple geometries [84], sheath dis­

tribution for simple, symmetric geometries [84], and equipotentiai lines [79]. One of the

studies provided ion energy and angle distributions [87]. However, none of the simulations

were capable of providing all of the relevant and important information for PSII, such as

matrix sheath distribution, Child-Langmuir law sheath evolution, electromagnetic field dis­

tribution, equipotentiai lines distribution, ion trajectories across the sheath, and the ion flux

distribution. Because of their choices of computational methods or the simplifying assump­

tions made in these studies, it was not possible to provide information for more complex

2D or 3D geometries.

PSII simulation using the MAGIC Code is presented in this Chapter. MAGIC is an

electromagnetic particle-in-cell, finite difference, time-domain code that solves Maxwell’s

equations and Lorentz force equations. The user can choose 2-D or 3-D geometry accord­

ing to the needs of the simulation. We found it successful in providing crucial information

for PSII, and much easier to use because of its ability to easily handle different geometries

and because the outputs are user-configurable. The plasma tool box eliminates the need for

detailed low level programming. Details are presented in the following sections.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2 Modeling of PSII Process 92

4.2 Modeling of PSII Process

4.2.1 Collisionless Model of PSII Process

Analytical models of planar collisionless sheath for the plasma source ion implantation

process [12,84] have been widely used to help understand the basic plasma physics process,

such as sheath formation and evolution. Such one-dimensional analytical models predict

the sheath evolution and implantation current density as a function of time. To accelerate

ions in the plasma to sufficiently high energy and to achieve narrow energy distribution, low

processing pressure is usually used for PSII. Here a quick introduction to the collisionless

model is given, which will be useful later to interpret the modeling results from MAGIC.

The planar PSII geometry is shown in Fig. 4.1. At time t = 0, a voltage pulse of

amplitude -Vo is applied to the target, which drives away the electrons in the plasma to

form a matrix sheath, which has a width determined by the amplitude of the voltage and

the plasma characteristics:

S„ = ( ^ V (Eq. 4.1) en0 It is assumed that the ion flow is collisionless and the electron motion is inertia-less.

The applied voltage Vo is much greater than the electron temperature Te, so that the Debye

length Xpe < so and the sheath edge at s is sharp.

A quasistatic Child Law sheath forms as the ions in the matrix sheath are being im­

planted [12,84]. The current demanded by this sheath is supplied by uncovering of ions at

the moving sheath edge and by the drift of ions toward the target. During the motion of an

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2 Modeling of PSII Process

Ion iJ|89UEi Matrixividirix Sheath

Figure 4.1. Planar geometry showing the formation of ion matrix sheath.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2 Modeling of PSII Process 94

ion across the sheath, it is also assumed that the electric field is frozen at its initial value,

independent of time, except for the change in field due to the velocity of the moving sheath.

These two assumptions permit an analytical solution be obtained for the sheath movement

and are justified post-hoc by comparisons with numerical results. The steady state Child

law sheath thickness is

sc = s0(=— )l/2 (Eq. 4.2) 9 uB

where u0 = (2eVo/M)1'2 anduB = [eTJM fl2.

The sheath thickness can also be expressed as

s = s0((2/3)u;pjt + l) 1/3 (Eq. 4.3)

where

^ = Q ' /2

This result (Eq. 4.3) can be obtained by assuming that the drift velocity of ions

toward the target is zero [84]. This assumption is also supported by experimental results of

time-resolved Langmuir probe measurements, showing the presheath was suppressed for

supersonic moving sheaths [89]. If the ions are assumed to have a drift speed of uB toward

the target, the same expression can also be obtained [12].

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3 Introduction to MAGIC Code 95

4.2.2 Modeling of PSII Process

Plasma simulation computes the interaction of charged particles with electrical and mag­

netic fields. In principle, direct integration of the Maxwell’s equations and the Lorentz force

equations should solve any electromagnetic problem and determine its dynamic behavior.

Most often, a plasma exhibits complex nonlinear behavior and the field and particles in­

teract with time-dependent boundary conditions. Usually, there are billions or perhaps

trillions of particles per cubic centimeter for even a low-density plasma, which makes the

simulation very computationally intensive [90].

Historically, computational plasma physicists wrote codes in structured languages

such as Fortran and C. These codes often have 104 to 106 lines of code not including the

user interface [90]. These ’’legacy” codes can often trace their origin to the 1960s or 1970s.

They were programmed to solve specific and specialized problems, requiring experts to run

and maintain them. Such codes are difficult to upgrade because of the close coupling of

their structure and solution techniques, thus modular upgrading is difficult or impossible.

Many times, the code developer was the only one who could run and maintain these codes.

4.3 Introduction to MAGIC Code

MAGIC Code is an electromagnetic particle-in-cell, finite difference, time-domain code

that solves Maxwell’s equations and the Lorentz force equations. Maxwell’s time-dependent

equations are solved to obtain electromagnetic fields and the Lorentz equation is solved to

obtain particle trajectories. Meanwhile, the continuity equation is solved to provide current

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 blAGIC Modeling of PSII Process 96

and charge densities for Maxwell’s equations [91]. MAGIC successfully integrated many

useful PIC techniques, enabling the user to address different geometries and boundaries,

materials, particle emission models, to interpret voluminous and abstract modeling output,

and sometimes to integrate with other software in a design environment. Consequently, it

has been used for the modeling in many areas, including antennas, sensors, Lasers, beam

propagation, plasma heating, ion sources, and field emitter arrays [92].

MAGIC works like a tool-box for the user to configure the proper system geome­

try, materials combination, boundary conditions, field algorithms, particle algorithms, and

output requirements. It represents the fields on a finite-difference grid. Particles are rep­

resented using the particle-in-cell approach, representing a large number of physical parti­

cles of the same species by virtual particles or macro particles. A variety of algorithms for

particle kinematics and current density allocation are individually optimized for different

density, velocity, and field strength [92].

4.4 MAGIC Modeling of PSII Process

Usually, large time steps and a small number of particles are desirable to reduce simulation

costs. However, the choice of particle number is limited by the spatial resolution needed

for the simulation. For the ion implantation process we are studying, the actual pulse width

is -10 fis. To show the trend and proper statistics for the physics, we have to opt for a long

simulation time. However, when simulating the implantation of small geometric features

on a large piece of substrate in a large-scale implantation chamber, using a large virtual par-

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 97

tide numbers and long simulation times makes the modeling computationally intensive. To

improve the spatial resolution of the simulation for small geometric features, it is important

to choose the finest possible grid for the space around those features. However, the com­

bination of these demands (fine grids, large virtual particle numbers, and long simulation

times) can cause the simulation to crash. Successful execution of simulation require care­

ful consideration of the desirable resolution, proper statistics, and acceptable computational

cost. Most of the time, a compromise of these factors is required.

In the simulations reported here, the ion plasma density was assumed to be 1 x

1010/cm3. The initial ion and electron distributions were assumed to be uniform. The

electron temperature was taken to be 4 eV [12]. The positive ions were singly ionized

nitrogen molecules (see Chapter 3 on RFI source designing and characterization). The pos­

itive ions were created cold (7] < 0.5 eV) and then accelerated by a high voltage negative

pulse of 100 KV. Using a discrete time step of 12.3 ps, the pulse was applied at f0 = 2 ns,

with a linear rise time of 0.1 ns. For all simulations with a large implantation chamber, the

HV pulse was applied to the sample stage using the PORT command, from an opening at

the bottom of the chamber. The simulated targets were placed on a 5 in. i.d. and 1 in. thick

conducting plate, supported by, and insulated, from the chamber bottom by a hollow quartz

ring (see MAGIC output graphs in following sections for more details).

The chamber and the sample stage geometry included in the models are assumed to

be azimuthally symmetric. Thus only a 2-D half-plane cut is necessary for full simulation.

The chamber simulated is 0.71 m ID, 0.61 m tall.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 98

4.4.1 Potential Structure, Field Distribution, Ion and Electron Distribution

MAGIC modeling results of several geometries are presented in this section, providing

information about sheath evolution, field distribution, and equipotentiai distribution. These

results are compared to other modelings to show the advantages of MAGIC modeling.

Field distribution and/or equipotentiai lines determine the ion trajectory across the

sheath, thus determine the implantation energy, incident angle, sputtering rate, dose distri­

bution, and the retained dose on the target. Sheridan, et al., calculated the ion matrix sheath

around a square bar solving Poisson’s equations and obtained equipotentiai and field lines

[79], which closely follow those found for a cylindrical electrode. The equipotentiai lines

followed the structure of the bigger square bar better.

We modeled the sheath evolution around a long cylinder sitting on the sample stage

as a benchmark. The electron distribution at T = 110.070 ns is illustrated in Fig. 4.2,

showing a well-defined sheath edge after the pulse rise-time transients have decayed [80].

The sheath surrounding the 1 in. i.d., 11 in. long conducting cylinder is greatly influenced

by the presence of the sample stage. The sheath around the middle of the cylinder is

almost cylindrical. However, the sample stage pushes the sheath away from the bottom

of the cylinder, resulting in larger sheath width. The sheath surrounding the top of the

cylinder resembles a half dome. The sheath edges are rounded (not showing sharp comers)

and follow the contours of the target assembly. The sheath width slightly decreases as it

approaches the top of the cylinder. This behavior is consistent with the results of previous

models [79].

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 99

Time 110.070 n s: PHASBSPACE fo r ELECTRON Q/M R atio: -1.759E+11

Z (M) IB-i)

Figure 4.2. The plasma sheath distribution at T = 110.070 ns for a negative voltage starting at Ta = 2 ns. The long cylindrical target was sitting on the sample stage, at the center of the cylindrial chamber.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 klAGIC Modeling of PSII Process 100

The field distribution for the same setup is shown in Fig. 4.3. The length of the

arrows are proportional to the field strength. The electric field converges at the top comer

of the cylinder and at the edges of the sample stage, confirming previous models [79] and

previous experimental measurements [82]. The electric field is weaker where the cylinder

meets the stage. The electric field is perpendicular to most of the cylinder surface. The

equipotentiai lines are also shown in Fig. 4.4. Equipotentiai lines are more rounded as the

distance of the line from the cylinder increases. The equipotentiai lines are parallel to the

side of cylinder, except at the top edge, and at the comer where the cylinder and the stage

join. The electric field is weaker in areas with sparse equipotentiai lines.

The positive ion distribution at To = 105.066 ns is shown in Fig. 4.5. The ion flow

is substantially laminar along the side of the cylinder. Ion focusing is apparent for the top

of the cylinder. The ion flux is not focused to the edge of the cylinder top, but rather to two

annular zones close to the edge, one on the top of the cylinder and the other on the side of

the cylinder. This effect was observed in previous experiments [82], and has been obtained

by previous PIC models [79,80]. The two smaller surfaces receive higher flux because of

the more focused electrical fields there. Similar results have been reported when the dose

distribution on a wedge-shaped implanted target was analyzed [82].

The electron and ion trajectories were also studied by making MAGIC ’’movies”,

showing the electron/ion distributions, field distribution, and equipotentiai lines distribution

at a series of time intervals, to study the trend of the process and to decide when a steady

state is reached for the implantation. By watching the movie made for the implantation of

the long cylinder, we found that the ions strike the edge of the cylinder top and the edge of

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 IE-11 ns. '-W y X /// = 100.062 T Z Z (Ml :::::::::::::::::

Time Time 100.062 ns: Vector plot of (Ez,Erho) 4.4 4.4 Process PSII of Modeling MAGIC

Figure 4.3. Electrical field distribution at 4.3. distribution Figure field Electrical o m CN tt- a ) Reproduced with permission ofthe copyright owner. Further reproduction prohibited without permission. 102 6.25 3.75 3.75 8.75 ! x B+4 11.25 13.75 16.25 I V/m), I 1 Ez -- : = = -i: (B-l) mm ns. 105.066 105.066 T =

Time Time 105.066 ns: X l-In te g ral o f Ez (V/m) 4.4 4.4 Process PSII of Modeling klAGIC vt v v v'v v v vt

Figure 4.4. Equipotential lines at4.4. lines Figure Equipotential V V V V V V V I t - a ) Reproduced with permission ofthe copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 103

Time 105.066 PRASESPACE f o r NITROGEN® LECULE Q/M R a tio : 3.446E+06

/jessA /A F SJSj'jr///////

8

;S?/: //////A

Figure 4.5. Plasma ion distribution at T = 105.066 ns, showing laminar structure near the side of the long cylinder and focusing near the edges.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 104

the sample stage at shallow angles, which causes more sputtering of the surface, shallower

implantation depth, and a lower retained implantation dose [12], even though the incoming

fluxes to these areas are higher.

This information can be easily extracted from calculations for different geometries.

We have modeled a thin plate, 23 in. i.d. and 0.5 in. thick, sitting on the stage. The results

for electrical field distribution, equipotential lines, electron distribution, and nitrogen ion

distribution are shown in Fig. 4.6, Fig. 4.7, Fig. 4.8, and Fig. 4.9. As shown in Fig. 4.6,

the electric field is very uniform and is perpendicular to the plate surface, except for areas

near the edge, where the field converges to the side of the plate. The apparent laminar flow

of the nitrogen ions is seen in Fig. 4.9 for the plate top and for most of the bottom side.

There are three annular zones near the plate edge which receive more ion flux, one on the

side, one on the top, and one on the bottom, caused by the local convergence of the electric

field. As mentioned above, the equipotential lines with smaller distance to the plate and/or

the stage follow the surface profile of the target assembly more closely (Fig. 4.7). The

sheath is more uniform for the plate top and is parallel to the plate surface.

Another geometry modeled and shown here is a 23 in.dia., 4.5 in. thick plate with

a 2 in.dia., and 4 in. deep hole in the center. Only the electron and ion distributions are

shown here (Fig. 4.10 and Fig. 4.11). There is apparent overlapping of the sheath with

chamber wall. Compared to the 0.5 in. thick plate above, there is less ion flux focusing

since the plate is thicker and therefore presents a larger surface area [82]. The laminar ion

flow is obvious for the top, the bottom, and the side of the plate. The sheath width at the

edge of the plate is smaller. In addition, there are some interesting focusing effects shown

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 105

Tine 109.947 osr Vector plot o£ (Ez,Erho) T7 7i 1------1------1------1------1------n M

O

Z (Ml (E-l)

Figure 4.6. Electric field distribution at T = 109.947 ns.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 106

Tine 109.947 nst Xl-Inteqral of Ez (V/n>

U

Ez (V/kJ, x E+4

1.25

13.75

21.25 Z (E-l)

Figure 4.7. Equipotential lines distribution at T = 109.947 ns.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process

Tine 109.947 as: PHASE SPACE for EIKTR01I Q/H Ratio: -1.759E+11

-i y * ' 1 * * A if

//.I • v.v-5

Ik&ZEtizzzzn l (M)

Figure 4.8. Electron distribution at T = 109.947 ns.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 hlAGIC Modeling of PSII Process 108

Tine 109.347 as: PHASESPACE £or lrrTROGElWOLEOJLE Q/M Ratio: 3.446E+06

Z (M) IE-1)

Figure 4.9. Nitrogen ion distribution at T = 109.947 ns.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 109

near the side wail of the center hole, which will be discussed in more detail in next section.

It is seen that the obtained sheath distribution (Fig. 4.10) is very similar to that obtained in

previous studies of a similar geometry using PIC codes [80].

According to Eq. 4.3, the sheath width for planar geometry at T = 109.947 ns

s = s0((2/3)wpii + 1)1/3 = 0.047 m (Eq. 4.5)

The observed sheath widths for the top of the thin plate was 0.075 m at T = 109.947

ns. The same value was obtained for the 23 in.dia., 4.5 in. thick, plate with a center hole.

The sheath width surrounding the side of the long cylinder (Fig. 4.2, uniform part) was 0.06

m. These snapshots of sheath distribution were taken at slightly different times, however,

the effect on the sheath width is negligible. The results show that a thickness difference of

3 inches does not have significant influence on the sheath width on the plate top. The thin

plate thus can be considered to have a planar geometry since its thickness is much smaller

than its diameter. However, the sheath width for plate top is much larger than that predicted

by planar analytical model. The sheath width for the side of the long cylinder is smaller

than that for the thin plate. Both are larger than the prediction of the analytic model for

planar geometry.

4.4.2 Ion Flux Distribution

The ion flux distribution inside a 2 in. diameter center hole in a 20 in. diameter cylinder

was modeled, using the option provided by MAGIC to monitor ion flux at any location

in the simulated space. A simulation time of 2.5 x 10~7s was used for each hole depth,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 ktAGIC Modeling of PSII Process 110

Time 109.947 as: PHASE SPACE f o r EIKTROII Q/M Ratio: -1.759E+11

'/A

0 1 2 3 4 5 6 Z IM) (E-l)

Figure 4.10. Sheath distribution at T = 109.947 ns.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 111

Tine 109.947 ns: PHASE SPACE for HITHOGEIIMOIECULE Q/M Ratio: 3.446E+06

Z (M) ( E - l)

Figure 4.11. Nitrogen ion distribution at T = 109.947 ns.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 112

representing the initiation of a quasi-steady state lasting throughout the pulse. The ion flux

on the side wall and on the bottom of the hole was monitored and the accumulated total ion

fluxes were output as functions of spatial location. The definitions of hole depth and radial

position are shown in Fig. 4.12. Three different hole depths were used for comparison. Fig.

4.13 indicates that the bottom of the deeper holes received much less ion flux, although

with better spatial uniformity. The ion flux distributions for the walls are shown in Fig.

4.14. The ion flux focusing effect found in Fig. 4.11 is confirmed by the maximum of the

ion flux distributions, located 4 cm away from the opening of both the 4-inch deep hole

and the 8-inch deep hole. The accumulated total flux for the wall of the 8-inch deep hole is

larger than that of the 4-inch deep hole, probably from reflected ions that could not reach

the bottom, where the total flux is smaller than that of the 4-inch deep hole. On the other

hand, the focusing effect is not found for the 2-inch deep hole, which has the highest total

flux for the bottom, due to the lower depth to radius aspect ratio.

The cases for side walls joining the bottom at a angle less than 90 degrees are also

modeled for dose distributions. The depth is 2 in. for both. The radii of the bottom are 1

in. and 1.5 in. respectively. The angle of the side wall is 45 degrees for the 1-inch hole

and 53 degrees for the 1.5-inch hole. The locations for flux monitoring are shown in Fig.

4.15. The flux distributions for the bottom with a 2.5 x 10_7s modeling time are shown

in Fig. 4.16, indicating higher fluxes for the centers and for the 1.5-inch hole. The flux

distributions for the side walls are shown in Fig. 4.17, both decreasing monotonically from

the top to the bottom. The bottom of the 53 degree wall received much less flux than that of

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 MAGIC Modeling of PSII Process 113

Depth

0 r

Figure 4.12. The definations of the radial location and depth for a hole for ion flux monitoring.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced Figure 4.13. Accumulated total ion flux Accumulated total Figure holes.4.13. distributions for the the bottoms of Flux (10 C/m ) 4 6 2 3 5 0 1 . 02 . 06 . 10 . 14 . 18 . 22 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 . h'LAGlC Modeling4.4 of PSIl Process Radial Position (cm) Position Radial 8-inchdeep hole 4-inchdeep hole 2-inchdeep hole 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced Figure 4.14. The accumulated totalaccumulated ion The fluxFigure 4.14. distributionsholes.the three on walls of

Accumulated Total Ion Flux (10'5C/m2) 2 3 0 1 0 2 in. deep in. 2 . MAGIC Modeling of PSII Process4.4 5 4 in. deep 8 in. deep in. 8 deep in. 4 Depth (cm)Depth 10 15 20 115 4.5 Summary 116

the 45 degree wall. These results are consistent with the TEM results of a implanted trench

presented in Chapter 5.

4.5 Summary

The MAGIC Code was successfully applied to model PSII process. With the many suc­

cessfully integrated PIC techniques of MAGIC Code, the modeling addressed different

geometries and boundaries, materials, plasma parameters, illustrated sheath formation and

evolution, field distribution, ion and electron trajectories, for critical informations, such as

ion incident angles and dose distribution, which are important for PSII experiment design

and for the understanding of the PSII process. The results presented here compared favor­

ably with experimental results and simulations from previous modelings by other groups.

MAGIC modeling is advantageous compared to l-D or pseudo 2-D analytical models, for

its 3-D and user-configurable multiple information output capability. It is also more con­

venient and more efficient to perform than working on low-level PIC code development or

modification of legacy code of plasma physics modeling.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.5 Summary 117

Location on the wall

0 r

Figure 4.1 S. Locations defined for monitoring accumulated total ion flux distribution for a trench.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.5 Summary 118

10

E 9 wO 1.5-inch O 8 X 3 LL C 7 o m o I- 6 "O 03 ra 3 5 E O3 o 4 < I-inch 3 0 1 2 3 4 Radial Position (cm)

Figure 4.16. The radial distributions of the accumulated total ion flux at the bottom of 2-inch deep holes

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Accumulated Total Ion Flux (10‘5C/m2) Figure 4.17. Accumulated totalionFigure 4.17.flux distribution on the trench walls. 10 9 4 6 8 5 7 3 0 Wall Joining the Bottom the Joining Wall at 53 Degrees 53 at 2 LocationSide Wallon the (cm) . Summary 4.5 Bottom at 45 Degrees 45 at Bottom Wall Joining the Joining Wall 4 6 8 119 Chapter 5

Pattern Writing by Implantation in a PSII System

5.1 Present Direct Write Technology

Direct write technology is under intensive study as one approach to integrate 3-dimensional

mesoscopic electronic components in a conformal manner, at low-substrate temperatures,

on a large variety of substrates. Mesoscopic electronic devices are defined as electronic

devices that have sizes ranging between conventional microelectronic devices (sub-micron

range) and traditional surface mount components (10 mm range). By integrating advanced

electronics, functional devices, and active components into the substrate, one can lower the

system weight and reduce cost [21]. The advanced electronic devices being considered in­

clude resistors, capacitors, inductors, high gain antenna, interconnects. Active components

include batteries. Functional devices include embedded sensors for pressure, voltage, tem­

perature, humidity, etc. The ability to directly-write power devices offers many system

advantages, from higher battery specific energy densities to lower overall system weight.

If the electronics and the power source are integrated, there will be no need for traditional

120

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 Pattern Writing by Implantation in a PSII System 121

printed board circuits. Ultimately, machines and electronics can become simple integrated

systems. Possible applications include miniature passive components for microsatellites,

munitions, toys, and wireless communications such as cell phones and GPS, embedded

sensors for condition based maintenance and for the medical industry.

Direct write also presents the possibility of replacing photoresist for all electronic ap­

plications by removing the mask step and reducing the number of steps required to produce

electronic circuits [93]. The ability of direct write manufacturable device on the substrate

will save money and process time. Current technologies used for direct write can be clas­

sified into two categories: deposition-based and print-based.

Deposition based technologies include thermal spray [94,95], molten metal droplet

deposition [96], micropen [97], focused ion beam [98], laser assisted chemical vapor depo­

sition [99] and ink-jetting [100]. Thermal spray processing techniques are of great interest

for large area coating in industrial applications. Thermal spray uses flame, arc, detonation,

and plasma for spraying the materials to be deposited [4]. In each case, powder precur­

sors are fed into the spray head where the local temperature is often in excess of 3000° C.

The heated precursor materials are subsequently accelerated onto the substrate at 90 m /s

for flame spray to over 700 m /s for detonating spray. The high temperature and high ki­

netic energy of the arriving particles are beneficial for the formation of adherent and dense

coatings but sometimes harmful to sensitive substrates. Fine tuning of the composition and

structure of the coating is difficult. Molten metal droplets also are not suitable for sensitive

substrates.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 Pattern Writing by Implantation in a PSII System 122

Variations of chemical vapor deposition (CVD) process rely on chemical reactions

on the substrate surface to produce the desired film, and thus allow reproducible film com­

position However, it also makes producing films with composition that vary from equilib­

rium difficult. CVD usually requires elevated substrate temperature to promote the desired

chemical reaction on the surface and to achieve good deposition rate, which may be un­

acceptable for some applications that need to keep the bulk material microstructure, or

materials that can not be heated to high temperature, such as most plastics or polymers.

Print-based or ink-based technologies use existing or modified print heads and special

’’ink” to print microscopic features on the substrates. Thus for every material that needs to

be direct written, a solution or suspension of the material with the proper viscosity has to be

found. After printing, drying or annealing are usually involved to obtain the final features

with right composition and properties.

Both deposition-based and ink-based technologies produced features on top of the

substrate surface. Both the original surface finish and the substrate size are therefore

changed by the writing process. There is no effective means to control the surface fin­

ish of the produced features, except perhaps by chemical-mechanical polishing after the

write step. For deposition based technologies, film adhesion is always an issue and surface

pretreatments or buffer layers may be found to be necessary. For ink-based technologies,

the proper solvent may not be available for the material that needs to be written. Even with

the materials can be written with ink, dense and uniform features are not easy to obtain.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 123

5.2 Pattern Writing by Implantation

5.2.1 Strategy and Advantages

High energy ion implantation is an important surface modification technique, which causes

minimal dimensional change, and avoids thermal distortion of the surface profile of the

treated object [11-13]. It also keeps the surface finish of the substrate. As a low-temperature,

surface processing technique, it changes the surface material properties without changing

the bulk materials microstructure and properties. Unlike thin film deposition, ion implan­

tation happens beneath the surface of the substrate. Therefore, there is no worry about the

bond strength of the converted layer leading to coating delamination. PSII is also a non­

equilibrium process; new alloys are not limited by classical thermodynamics properties and

diffusion kinetics. By control of implantation time and voltage, graded surface layers can

easily be produced.

Plasma source immersion ion implantation uses the plasma of the species to be im­

planted as the ion source. The target to be implanted is usually immersed in the plasma and

all the exposed surfaces of the target are implanted simultaneously when a train of high

voltage negative pulses is applied to the sample stage. Compared to beam line implanters,

It is advantageous for large-area, high-dose, and three-dimensional target implantations.

Our planar RFI source for the large-scale PSII system has been found to be able

to provide high density plasma with good spatial uniformity. It is capable of large area

uniform processing. It is also feasible to selectively convert surface area properties of very

large substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 124

In this study, pattern writing has been performed by exposing metal and semicon­

ductor samples through masks, to high ion doses. It was studied as an ink-less, deposition

free, mask-based direct write technique as an alternative to current direct write technolo­

gies. AFM calibration gratings, made from Si, were also implanted. Implantation depth

profiles were characterized by variable-angle spectroscopic ellipsometry (VASE) and AES

and compared to TEM results. Measured lateral and depth profiles are compared to the

feature morphology to assess lateral diffusion, pattern transfer fidelity, and wall-effects on

the depth profile. This chapter also presents the results of MAGIC modeling of flux and

angle of ion trajectories through the boundary layer predicting the magnitude of flux as a

function of 3-D location on objects in the expanding sheath, and to evaluate the fidelity of

pattern transfer as a function of feature size.

5.2.2 Experimental Conditions

Two kinds of substrates were used for studying pattern transfer: flat substrates and silicon

gratings. Flat substrates, including silicon wafers (100 mm in diameter, 0.5 mm thick),

polished stainless steel plate (9.12 cm in diameter, 1.2 cm thick), and pure Ti and Ta sheets

(25 mmx25 mm, 0.5 mm thick for Ta, 1 mm thick for Ti), were implanted through a

150 pm thick metal mask with 100pm wide lines and 100 pm ID holes. The mask was

weighted to secure it to the stage using a shaped metal ring and a metal cube. The ring is

2.5 cm tall, with a 2.5 cm OD and 2.3 cm ID. The cube is 3.8 cmx3.8 cmx3.8 cm. The

ring and the cube were carefully positioned on the 100-mm Si wafer to study wall effect on

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 125

dose and sheath distributions. Pure silicon gratings with trapezoidal groove cross-sections

and 10 pm groove separation were also implanted.

Research grade nitrogen (99.999%) was used as the gas source. A base pressure of

~-10-8 Torr and operating pressures of 5 x 10~4 to 2.5 x 10~3 Torr were used. Silicon

wafer implants were performed at pulse voltages of 50 kV and repetition rate of 20 Hz for

10 hrs. Stainless steel was implanted at 50 kV at 25 Hz for 9 hrs. The Ti and Ta sheets

were implanted through the mask at 40 kV at 25 Hz for 7.5 hrs. Si gratings were implanted

at 25 kV at 40 Hz for 5 hrs. The estimated temperature rise for these implantation runs was

less than 50°C.

AES and XPS spectra were taken using a VG Scientific MARKIIESCALAB equipped

with EX05 ion gun for depth profiling. Nitrogen concentration depth profiles for high dose

nitrogen implanted Si and Ti were obtained by AES. The sputtered depths were calibrated

by surface profilometry. The overlapping of nitrogen and titanium peaks and compensation

for the presence of oxygen were considered when calculating the nitrogen concentration

[101]. The nitrogen AES peak is overlapped with titanium 385 eV peak. Usually, the con­

centration of nitrogen is deduced by a subtraction considering the ratio between the two

titanium AES peaks, namely the 385 eV peak and the 418 eV peak. However, when there

is oxygen in the system, things become more complicated because this ratio changes with

different oxygen concentration. To reach the right concentration of implanted nitrogen for

our samples, the influence of oxygen was also considered following methods developed by

previous studies [101].

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 126

A variable-angle spectroscopic ellipsometer (VASE) [50] was used to study the opti­

cal properties of the nitrogen-implanted silicon and stainless steel samples. Three incident

optical angles (70,75, and 80 degrees) were used for each area. The incident optical beam

was about 3 mm in diameter. Because the as-received Ti and Ta samples did not have

smooth optical surfaces, ellipsometry could not be used to examine them.

The cross-section of nitrogen implanted Si gratings was examined by TEM at the

Solid State Division of Oak Ridge National Laboratory.

The depth profiles for high dose implantation were also calculated using Profile Code

[102].

5.2.3 Results and Discussion

The implanted Ti surface showed the gold color characteristic of stoichiometric TiN, with

a slight pink hue because of the slightly oxidized surface layer. The AES depth profile for

high dose nitrogen implantation of Ti is shown in Fig. 5.1 for a Ti substrate. The nitrogen

concentration fluctuates around the level for stoichiometric TiN (50%) throughout most of

the implanted depth, decreasing sharply at -5000 A. The native oxide layer of the original

substrate was not stripped prior to implantation, accounting for the relatively low nitrogen

concentration in this range. Similar nitrogen depth profiles were obtained for the implanted

Si wafer (Fig. 5.2), again showing a thick native oxide layer that reduced the nitrogen

reaction with silicon, producing a low nitrogen concentration.

The experimental implantation depth profiles are in good agreement with the high-

dose model calculation of Profile Code, which predicts both the constant concentration for

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 127

most of the implanted depth and the steep decrease at the end of range [102]. As shown

in Fig. 5.3, the nitrogen concentration is saturated for some range at an implantation dose

of 1.0 x 10 18/cm 2. As the dose is further increased, the little valley of the concentration

curve is further filled to the saturation level, to result in a very flat depth profile (Fig. 5.4).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 128

AES Depth Profile of Nitrogen Implanted Titanium

60 - C O)0) o 50 - •••*•• V*vV Z i • • »• o 40 • OJ

0 1000 2000 3000 4000 5000 6000 7000 Depth from the Surface (angstroms)

Figure 5.1. Implanted nitrogen concentration depth profile for Ti substrate obtained by AES depth profiling

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 129

Test Position Thickness of Layer (A) Color of Spot A 1770 ± 2 0 Golden B 914 ± 3 Blue C 1055 ±3 Light Blue D 1580 ±20 Golden (lighter) E 851 ± 2 Blue F 1620 ± 2 0 Golden (lighter) G 1820 ± 2 0 Golden

Table 5.1. Nitrogen implanted layer thickness on Si substrate obtained by ellipsometry measurements

The ellipsometry results for nitrogen implanted Si and stainless steel samples con­

firmed the film-like depth profiles observed by AES and predicted by Profile Code. Table

5.1 shows VASE results for 7 different spots on implanted Si wafer. Adding a graded opti­

cal layer to model the gradual concentration changes typical of low-dose implantation depth

profile, did not improve the fits to the data. The optical properties of the implanted layer

closely resemble that of an extremely uniform thin film. These optical methods thus provide

a dose distribution for high dose implantation, since the dose is in proportion to converted

layer thickness and thickness can be correlated to area color. They are non-destructive and

quick compared to other depth profiling techniques, such as AES and SIMS. The color of

100 pm lines and 100 pm ID dots produced by implanting through the metal mask were

studied using an optical microscope fitted with a CCD color camera and TV monitor. The

edges between the implanted and un-implanted area were narrower than the resolution of

our camera, indicating pattern transfer contrast considerably sharper than 2 pm. A 5.7 pm

wide, blue-colored annular zone appears along the circumference of the exposed area. El­

lipsometry indicated that this annulus received only -40% of the maximum dose received

at the center of a 100 pm hole. The sample stage temperature did not rise significantly dur-

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 130

AES Depth Profile of Nitrogen Implanted Silicon

O 30 -

1000 2000 3000 4000 5000 6000

Depth from Wafer Surface (angstroms)

Figure 5.2. Implanted nitrogen concentration depth profile for Si substrate obtained by AES depth profiling

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 131

70

60

50

40

30 o> 20

10

0 0 500 1000 1500 2000 2500 3000 3500 Depth (Angstroms)

Figure 5.3. Nitrogen concentration in Ti substrate calculated by high dose model of Profile Code. The implantation voltage is 50 KV and the dose is 1.0 x 10 l8/cm 2.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 132

60 -

50 - c o S 40 - o0)

O 30 - C o>0) 2 20 - z

0 1000 2000 3000 4000 Depth (Angstroms)

Figure 5.4. Nitrogen concentration in Ti substrate calculated by high dose model of Profile Code. The implantation volatge is 50 KV and the dose is 2.5 x 10l8/cm 2.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 13 3

ing implantation. Analysis of the annular zone just outside the implanted circle confirmed

our assumption that lateral thermal diffusion may be ignored.

The AFM calibration gratings were exposed to study the dose distribution in and

around narrow features. The trapezoidal groove cross-sections provide virtual mask stripes

over the flat Si surface. Our apparatus may be operated in a manner which allows us

to co-deposit Si02. This mode is discussed in Chapter 3 on RFICP source designing and

characterization, and in Chapter 6, on nitrogen-doped Si02 films deposition. Fig. 5.5 shows

TEM picture of Si gratings implanted in this mode. As shown in Fig. 5.5, the implanted

layer is capped with co-deposited SiOo. The flat tops and bottoms of the grooves show

a similar implanted layer thickness. The sloped sidewalls were not uniformly implanted.

The converted layer near the comer where the bottom and the sidewall meet is apparently

thinner, while the layer near the comer where the top and the sidewall meet is much thicker.

We modeled the dose and angle distribution for mask-based pattern writing using

MAGIC, an electromagnetic particle-in-cell, finite-difference, time-domain code for simu­

lating plasma physics process [91], which we adapted for this purpose. The chamber and

sample stage geometry included in the models are assumed to be azimuthally symmetric.

Thus only a 2D half-plane cut is necessary for full simulation. We have used this simulation

to examine matrix sheath formation, sheath propagation, evolving to the steady Child-Law

sheath; ion trajectories, ion depletion, and the distribution of ion flux for different mask

shapes and aspect ratios. In the simulations shown here, the time step was 5 x 10-los over

a simulated implantation time of 2.5 x 10-Ts , representing the initiation of a quasi-steady

state lasting throughout the pulses. The modeled ion plasma density was 1010/cm 3 and ap-

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2 Pattern Writing by Implantation 134

S1O2 cap

Implanted Layer

Si Substrate

Figure 5.5. TEM picture of cross-section of nitrogen implanted Si gratings, showing the nitrogen-implanted layer, the Si substrate, and the Si02 cap layer

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3 Conclusion 135

plied voltage -10 kV. The positive ion distribution at 1.2 x 10~7s modeled for a groove

with a sidewall to bottom angle of 53.1 degrees is shown in Fig. 5.6. The shallower an­

gle of incidence at the slopes sidewall predicts a non-uniform distribution consistent with

TEM observations. The ion flux distribution (Fig. 5.7) predicts that the bottom of the slope

will receive a factor of 2 to 3 less flux, also in agreement with TEM results. Trenches with

sidewall joining the bottom at different angles were also modeled and the ion flux distrib­

utions were obtained. More details were given in Chapter 4 on MAGIC modeling of PSII

process. With carefully designed mask shape and thickness, the uniformity of implantation

and shorter implantation time can be achieved.

5.3 Conclusion

High dose nitrogen implantation into various substrates was performed, yielding film-like

depth profiles. A stoichiometric TiN layer was produced on the Ti surface. Depth pro­

files obtained experimentally by AES, VASE and those modeled with Profile Code are in

good agreement. Patterns 100 nm in line-width were transferred by implantation, with lat­

eral contrast better than 2 fj.m . The TEM study of nitrogen implanted Si gratings showed

implantation can produce sharp patterns as small as 3 fim with good uniformity, in agree­

ment with MAGIC code simulation. With carefully designed masks and good control of

implantation parameters, PSII can be used for pattern writing of smaller features.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3 Conclusion 136

'S S S S S S J 'S j l 'j 'j'S i 'S s y j 'J ’J J / J 'j 'j r j 'j 'j y f j s j ’S j t s j f j ’y s f j 'j f j r j/ 'S s j S 'S & S J 'S S S J 'S j r S j '/ i.ii M. It MI'IH'IIII milHIHaHI till ll'llll!" I! ll|l* IIIIIKI iiiiiiia itlttltil» l'|l'« »|ll||lflfm fiUWV''1 liiiitlill^flllllllqlillillillillltililiift niiiimiiiaii liiiiiiiumiiiiiiatii

i.i. imamiiiiiiiiiitiii.i^iiiuiiiii

mi Mi*ai!li«i in I'll'll'B !!! ilililiiiiil!! |l .[tiDtiDiiUMHjiitinhti.iiaiM .^IIII'llllllllltllllB I I'lilll.Hlill'HMIMIlilllltitlUt

i i m l l l i l i i t i i t i mm '•iiiiaHiniiihiMiiH H llll! iiiiiliiain il! !I i i i i i ||I > h i i i m h m iinniBi 'IIMMIIIII •MHMIMIII! HIM II iaiM(tiihl4tili •|!,'hia,"iiinuiMii itiiiiiitimiiBiiiiiiiiiimii l l ' i ' l | | 4 'HI h ...... • llll)|«l|ll|tll*lllllMtthll«M I » t l l l l l t II : i :i ini i! !l i • } : ■ f 1 / liiajiii$niM|j|| | | , ,,,-,1111 i.iuiw;'/// t MIII ■ 1111 ■! ’" III11 iili!!!!! f e a I I I I iiimllM iili immiiiii iirfniiiiiiuniiUil.i iiiwnti IH m ! i l l

Figure 5.6. MAGIC modeling of PSII process in a trench, showing positive nitrogen ion distribution at time 1.2 x 10~7s.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3 Conclusion 137

0 r

i

0 1 s 3

:§

U.3 I |

0 t i 4 9 t Location on ttto Sitfo Wo II (cm)

Figure 5.7. MAGIC simulation results of ion flux distribution along a trench surface. The trench is 2 in. deep, 3 in. wide at the bottom, with side wall joining the bottom at 53.1 degrees. The simulation time is 2.5 x 10~7 s.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 6

Deposition of N-doped Si02 Films by Novel Control of the rf Operation Mode

6.1 Introduction

High quality Si02 film is important for ULSI [103] and for optical Si-based waveguides

fabrications [104]. It has been produced by methods including various kinds of CVD [23,

103,105-107], thermal oxidation [103], oxygen ion implantation [104,108], aerosol-gel

deposition [109] and by Laser or electron beam ablation [110—112]. SiH4 process is popular

for CVD Si02 because of its low deposition temperature and high growth rate, however,

SiH,t is not safe to handle. Thermal oxidation and oxygen ion implantation require Si

substrate to generate Si02. Laser or electron beam ablation usually produces rough film

surfaces.

A small concentration of nitrogen in the Si02 film and at the Si-Si02 interface has

been found to improve the reliability of the dielectric film and to prevent doped boron

diffusion [3,105,106]. Spatially controlled nitrogen incorporation Si02 layers for device

138

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 Deposition o f N-doped SiOi Films by Novel Control of the rf Operation 139 Mode

manufacturing has been studied [22]. Hydrogen-free deposition gases have also been in­

vestigated for higher film quality [23,106]. As an alternative, Si02 films can also be doped

with F to achieve a lower dielectric constant for ULSI applications [113].

In this work, a novel operating regime has been used in our P S II apparatus which

we called PSII/PVD. Using this controllable regime, our system has been used to deposit

large-area, uniform, N-doped S i 0 2 films. This method shows promise for depositing doped

S iO oat very low temperatures, on a variety of substrates, allowing a range of dopant com­

binations to produce variable dielectric constant, and spatially controlled nitrogen incorpo­

ration into the deposited layers. The system uses planar rf coil for plasma generation and

for S iO o film sputter deposition. Nitrogen is incorporated into the deposited films and the

film/substrate interface by a combination of P S II and continuous plasma nitriding. High

quality nitrogen-doped S i 0 2 films have been successfully deposited on S i, Ti, glass, and

stainless steel substrates with excellent adhesion. Compared to other gases for S i 0 2 depo­

sition [103,106], N2 is safe to handle and is available in high purities at low cost. N2 is

also easy to pump and is benign to vacuum systems. In addition, the deposition plasmas is

inherently hydrogen free so the amount of hydrogen incorporated is minimized. The PSII

apparatus can also be employed to perform in-situ substrate surface cleaning and condi­

tioning prior to film deposition. In addition to S i substrates, S i 0 2 films can be produced on

other metal, glass, or polymer substrates.

Nitrogen-doped SiOo film deposition by the PSII/PVD system is discussed in this

chapter. The system design and operation is described. The deposited films were studied by

AES and by SIMS concentration depth profiling. Variable angle spectroscopic ellipsometry

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.2 Experiments 140

(VASE) also was employed to investigate film optical properties and thickness, which was

compared to and confirmed by surface profilometry results. The cross-section of processed

small features was studied by TEM to reveal the multi-layer structure and the degree to

which the process preserves small feature dimensions during processing.

6.2 Experiments

The PSII/PVD system is shown schematically in Fig. 6.1. It consists three sub-systems:

the vacuum deposition chamber, the rf plasma source, and the high voltage pulse forming

network. Detailed information of the PSII/PVD system is given in Chapter 7.

Three different kinds of substrates were used: pure titanium sheet (nominal purity

99.6%), silicon wafer (with a thin native oxide layer), silicon AFM calibration gratings

(with a thin native oxide layer), and highly polished stainless steel test electrodes for the

FEL at TJNAF. The titanium sheet is 2.54 mmx2.54 mm, 1 mm thick. The Si wafer used

is 100 mm ID, p-doped, and (100) oriented. The pure Si grating is 5 mmx5 mm and 1 mm

thick, with the center 3 mmx3 mm area consisting parallel trenches 2 pm deep and 2 pm

wide. The stainless steel test electrode is -0.15 m diameter, 0.019 m thick plate having a

Rogowski surface profile [2]. It was also polished to a 1 pm rms surface finish.

Research grade nitrogen (99.9995%) was used for all depositions. The system base

pressure before processing was better than 1 x 10-6 Torr. All the samples were placed on

the sample stage in the middle of the chamber without active cooling.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.2 Experiments 141

RF ftrwtrSifliy

GxiingWltr

Figh-tdtagc ftilse Farting I'iluak

FoaWiugt

MxftricJ RrtpB VbdnKa F\rrp Tiitoftitp A

Figure 6 .1 . The integrated P S I I /P V D system for large-area, uniform, doped S iO o films depostion.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.2 Experiments 142

The titanium sheet was first exposed to the plasma with an rf power of 360 watts and

nitrogen pressure of 5.2 x 10-4 Torr for 1.7 hr. The substrate was then pulsed at 38 KV

and 25 Hz for 3.7 hr. The high voltage pulse was 10 ps wide, with an average current of

21.3 A. Part of the titanium sheet was masked to create a untreated zone for comparison

purposes.

The silicon wafer was processed at 5.0 x 10“4 Torr with an rf power of 750 watts for

9 minutes, then pulsed at 30 KV at 25 Hz for 4.63 hr. A metal mask was fixed on part of

the wafer for film thickness measurements.

The silicon gratings were processed at a nitrogen gas pressure of 8.0 x 10-4 torr.

The rf power used for plasma generation and sputtering was 750 watts. The substrate was

processed in the plasma for 20 minutes before it was co-implanted at 25 KV and 40 Hz for

5 hr. with 20ps wide pulses. The average current during each pulse is 10 A.

The stainless steel test electrode was first processed with an rf power of 700 watts

and at a nitrogen pressure of 2.0 x 10-3 Torr for 6.15 hr. Then it was pulsed at 25 KV at

20 Hz for 2.58 hr., with an rf power of 750 watts and at a nitrogen pressure of 1.8 x 10-3

torr. The pulse width was 20 ps, with an average current of 10 A during each pulse.

The processed Si gratings, Si wafer, and the titanium sheet were studied by AES

for surface composition. The processed Si wafer and the titanium sheet were also depth

profiled by AES. The untreated zones of the Si wafer and the titanium sheet were also

depth profiled by AES for comparison. The processed Ti sheet was also studied by SIMS

depth profiling. The cross-section of the processed Si gratings was also studied by TEM

at Solid State Division of Oak Ridge National Laboratory. The Si02 film on Si wafer

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.3 Results and Discussion 143

was studied by VASE for its thickness and optical properties. The thickness obtained was

compared to the thickness as measured by surface profilometry (Dektak3ST surface profile

measurement system).

6.3 Results and Discussion

6.3.1 Operation of the PSII/PVD System

Planar coil rf plasma source is capable of producing large-area uniform plasma for mate­

rials processing at low pressures (<50 mTorr) [12,17]. Since rf power is coupled to the

plasma inside the vacuum chamber usually through a quartz window, there are no electri­

cal connections or supporting apparatus inside the chamber to take up extra chamber space

to complicate chamber geometry, which is important for 3-D uniform PSII of large tar­

get. Operated in inductive mode, it is clean and long lasting plasma generating source for

ion implantation, which requires low processing pressure for a long mean free path [114].

The planar rf coil is the inductive component of the rf impedance matching network, which

inevitably causes electrostatic coupling between the coil and the plasma. Electrostatic cou­

pling is important for the ignition of the plasma and helps to sustain the discharge at low

plasma ion densities. For the sub-mTorr pressure and low rf power (< l Kw), capacitive

coupling dominates and there exists large self-bias on the rf coil, which produces an rf

sheath at the plasma-window boundary [17,68,115]. This high voltage sheath accelerates

ions in the plasma to sputter the window material. In our system, we can choose this mode

to produce high purity SiOo films from our optical quality quartz window.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.3 Results and Discussion 144

For our preliminary studies, nitrogen was used as the processing gas for all of the

substrates. However, Ar can be used for sputter deposition if pure SiC >2 is desired. Dopants

other than nitrogen can be introduced into the system, either being mixed with Ar, with

other dopants, or being used as the sole low-pressure working gas. Stacked layers can also

be produced by introducing different dopant gases for controlled time periods during the

process. Implantation frequency can be adjusted during the deposition phase to produce

films with different dopant concentrations.

The rf source can be controlled to operate in an inductive mode without deposition.

This can be done by a combination of plasma parameter control and /or adjustable electro­

static shielding. In this mode, large-area uniform PSII can be performed to produce doped

layers of different thickness and implantation dose. This pure PSII mode can be used for

substrate sputter cleaning and conditioning, pre-deposition implantation, or post-deposition

implantation in the deposited layer. Patent applications have been prepared regarding these

novel and useful ideas.

6.3.2 AES and SIMS Depth Profiling

The AES depth profiles of the un-processed and processed Si wafer are shown in Fig. 6.2

and Fig. 6.3. As shown in Fig. 6.2, the bare Si sheet used has a thin oxide layer about

18 A thick. The processed Si wafer is covered by Si-oxynitride layer about 1700 A thick,

consisting 60% oxygen, 32% silicon, 7% nitrogen for the 50 A to 1700 A range. Because of

the incorporation of nitrogen, the oxygen concentration in the film is slightly lower than the

stoichiometric level. The carbon concentration for this range is below 1%. At the interface

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.3 Results and Discussion 145

of the deposited film and the substrate, there is a nitrogen-rich zone from 1750 A to 1840 A ,

where the nitrogen concentration is more than 10%. The peak concentration for nitrogen

is found to be 16% at 1790 A . The higher concentration of carbon and nitrogen and the

lower concentration of oxygen and silicon in the topmost 50 A range were attributed to the

surface absorbed layer on the film resulting from air handling. The existence of the doped

Si02 layer was also confirmed by SIMS depth profiling

The AES depth profiles of the un-processed and processed area of Ti substrates are

shown in Fig. 6.4 and Fig. 6.5. There exists an oxide layer on the bare substrate surface,

with oxygen concentration higher than 10% for the top 260 A range. The oxygen concen­

tration decreases to below 5% after 400 A . Carbon concentration varies around 4% level

throughout the profiled depth except the top layer, where higher carbon and Si concentra­

tion were detected. The average concentration of Si in the bare substrate is lower than 3%.

Depth profiling of the processed Ti substrate shows about 2250 A of Si-oxynitride layer,

consisting of about 60% oxygen, 30% silicon, 7-8% nitrogen, and less than 2% of carbon.

Carbon concentration is highest at the film-substrate interface again from air handling. The

oxygen and silicon concentration are very close to stoichiometric Si02. Depth profiling by

SIMS also confirmed the existence of around 2000 A thick silicon-rich layer (Fig. 6.6).

6.3.3 TEM Study of the Cross Section of Processed Si Gratings

The TEM picture of the cross-section of processed Si gratings is shown in Fig. 6.7. The top

of the grating is covered by a nitrogen-doped Si02 layer, which is about 4000 A thick for

the trench top and bottom, and about 2200 A thick for the trench sidewalls. As shown in the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Atomic Concentration (%) 100 0 ■40 60 - 70 - 90 0 2 -30 SO- - 80 0 1 U not implanted U - - 0 0 0 0 0 10 4 10 6 200 160 160 140 120 100 80 60 40 20 Figure 6.2. AES depth profile of AES profiledepth Figure 6.2.bare Siof substrate. . Results and Discussion 6.3 BigSi piacs Sputter Depth (A) 146 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Atomic Concentration (%) 100 Figure 6.3. AES depth profile of depth AESprofile Figure Sisubstrate.nitrogen-doped on 6.3. of coating Si02 Si02 sputter rate sputter Si02 N in piece large Si, o o s . Results and Discussion 6.3 1000 Sputter Depth 1500 (A) ■'v 2000 2500 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Atomic Concentration (%) 100 0 ■90 "! 0 6 70- 80- Ti, large piece - 40 - 50 , . . . 30 not implanted Figure 6.4. AESFigure depth 6.4. profile Tiof bare substrate. . Results and Discussion 6.3 Sputter Depth (/t) 7i2 600 i

800

1000

1200 1200 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Atomic Concentration (%) 100 0 t 50 Figure 6.5. AES depth Figure profilecoated 6.5.Tisubstrate. nitrogen-dopedof Si02 SI02 sputter rate sputter SI02 edge InTi,N 500 1000

. Results and Discussion 6.3 1500 - ot N1 sit 503000 2500 504000 3500 149 6.3 Results and Discussion 150

B M M aty. N Implanted TI, NagaUva iora

Si-»

SiN

10 Ti.Si SECONDARY ION INTENSITY (cts/sec) INTENSITY Ti.Si SECONDARY ION

18 10 100 0 1000 2000 3000 4000 5000 6000 APPROXIMATE DEPTH (A)

Figure 6.6. SIMS depth profile of nitrogen-doped SiC >2 coating on Ti substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.3 Results and Discussion 151

picture, the nitrogen implanted Si layer extends beyond the deposited film. The deposited

film was confirmed by AES to be a Si oxynitride layer. The AES study of the surface of

the film on the trench bottom shows 48.9% silicon, 27.6% oxygen, 13.3% nitrogen, and

10.3% carbon, whereas the concentration for the top is 49% silicon, 30.1% oxygen, 13.0%

nitrogen, and 8.0% carbon, nearly the same, and probably within the expected error from

the combination of the process variability and measurement methods.

6.3.4 Ellipsometry and Surface Profilometry Study

One area of the SiOo film deposited on the Si wafer through the metal mask was studied by

Ellipsometry for its optical properties and thickness. A typical result is shown in Fig. 6.8.

The deposited film is highly transparent, showing interference colors when viewed under

fluorescent room light. The measured index of refraction is 1.483 at 5086 A. The handbook

value for fused quartz is 1.462 [29].

The surface profilometry results also confirmed the thickness results obtained by el­

lipsometry studies. The result of one of the measurements is shown in Fig. 6.9. Considering

the processing time for the Si wafer, the average deposition rate is about 8-9 A/min. Be­

cause a metal block was used to clamp down the metal mask to the Si wafer, there are areas

of different film thickness available for profilometry study. The thickness of each area was

correlated to color appearance to investigate film thickness distribution.

The film deposited on the stainless steel FEL test electrode was inspected visually

for color distribution, to determine film uniformity [114,116]. The picture of the processed

electrode is shown in Fig. 6.10. The processed top surface is very uniform, showing

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.3 Results and Discussion 152

Si0 2 cap

Implanted Layer

Si Substrate

Figure 6.7. TEM picture of cross-section of processed Si AFM gratings, showing the nitrogen-implanted layer, the Si substrate, and the SiC >2 cap layer.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.3 Results and Discussion 153

Cauchy Optical Constants

1.52 0.0050

0.0040 -c 1.50 ~k c 3 o O O 0.0030 CD 3 1.48 o £ O o 0.0020'1 x o 0) <5‘ ■g 1.46 - 0.0010

1.44 0.0000 400 500600 700 800 Wavelength in nm

Area thickness: 2447 ± 9 A

Figure 6.8. Ellipsometry study of the thickness and optical properties of N-doped Si02 film on Si wafer. Cauchy optical model was used to model the experimental data. The obtained film thickness is 2447±9 angstroms. The film is almost transparent, with index of refraction of 1.483 at 5086 angstroms. The value for fused quartz is 1.462 at the same wavelength.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.3 Results and Discussion 154

Figure 6.9. Profilometry measurement of nitrogen-doped Si02 film thickness

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.4 Summary and Conclusion 155

the same interference color when viewed at a fixed angle. This indicates very uniform

film thickness. The curved edge at the sides of the top shows color stripes parallel to the

circumference of the plate, indicating different film thickness. This is to be expected, since

these areas face the quartz plate at different angles. HV breakdown test at TJNAF showed

that the nitrogen-doped SiC >2 film on a 0.15 m ID. FEL electrode has helped to suppress

field emission. The electrode was able to operate with a field gradient of 24 MV/m, with a

dark current of only 160 pA, after shorter conditioning time. The un-processed electrode

for the FEL is currently operated with a field gradient of 3 — 6 MV/m, with a dark current

of a few hundred nanoamps.

6.4 Summary and Conclusion

A novel operating method has been discovered in our large-scale PSII system and has been

used for large-area, uniform, doped SiC >2 films deposition. The PSII capability enables in

situ substrate surface cleaning and conditioning, ion implantation during film deposition,

pre- and post-deposition implantation to incorporate dopant into the processed substrate

with spatial control. Nitrogen doped Si02 films have been deposited on Si, Ti, and stainless

steel substrates. The concentration depth profiles obtained by AES and SIMS show that the

film is near stoichiometry SiC >2 film with 7-8% of nitrogen throughout most of the deposited

layer. Higher nitrogen concentration was found for the interface between the film and the

Si substrate. Ellipsometry results for one of the processed area indicate the film is highly

transparent with index of refraction close to the value for fused quartz. The obtained film

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.4 Summary and Conclusion 156

Figure 6.10. FEL stainless steel test electrode processed by PSII/PVD. The electrode is 0.15 m ID.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.4 Summary and Conclusion 157

thickness of the area is 2450A, which is confirmed by surface profilometry study. Large-

area Si02 film has also been deposited on 0.15 m ID highly polished SS test electrodes.

Except for the curved edge of the Rogowski surface profile, the film on the electrode surface

is highly uniform. HV breakdown test at TJNAF showed that the nitrogen-doped Si02 film

on a 0.15 m ID. FEL electrode has helped greatly to suppress field emission.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 7

Surface Processing by Plasma Source Immersion Ion Implantation

7.1 Introduction

Ion implantation is the process of accelerating ions into a material, changing its surface

chemical composition, and electrical, mechanical properties. Traditional ion implantation

is done by beam-line implanters. As shown in Fig. 7.1 [10], there is usually an ion source,

from which an ion beam is extracted. The beam is then accelerated and steered toward the

target. Because beam size is usually smaller than the target size, beam rastering is needed

for uniform implantation. When a three dimensional target is implanted, target tilting and

rotation are necessary to provide a uniform dose distribution. Plasma source immersion ion

implantation is advantageous for implanting large area, three-dimensional, and complicated

shaped target [10,13-15]. As shown in Fig. 7.2, the target is immersed in the plasma of

the species to be implanted. When a high voltage negative pulse is applied to the target, the

positive ions in the plasma are accelerated and implanted into the target. All the surfaces of

the target that are exposed to the plasma are implanted simultaneously. PSII is a non-Iine-of

158

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 Surface Processing by Plasma Source Immersion Ion Implantation 159

sight surface treatment technique. It is not limited by beam size or beam current, especially

suitable for high-dose, large area, and three-D target surface treatment.

PSII is significantly different from the conventional mass-separated and single-energy

beam line implantations [10-12]. For example, in an oxygen plasma, positive ion species

include O t and 0 +. For a nitrogen plasma, positive ion species include N f and N +. The

substance to be implanted is introduced into the chamber in gaseous form. Proper pressure

(usually in thelO-4 Torr range) is maintained for plasma generation. Plasma species con­

centration, ion energy distribution, plasma sheath distribution, and ion density distribution

are important factors determining dose distribution and depth profile for PSII. When the

negative voltage pulses are applied, all the positive ion species in a plasma are accelerated

and subsequently implanted.

Short duration, high voltage pulses are used for PSII. The pulses are typically 10—100

KV, 10 - 50 ps long, at a repetition rate around 100 Hz. During implantation, the positive

charges accumulated on the target surface need to be neutralized by the negative charges

from the plasma. At high implantation voltages, the ion induced secondary electron emis­

sion coefficient is around 4.90 for 30 KV and 6.33 for 50 KV [2], which aggravates the

charging problem. On the other hand, a sustained high voltage in the plasma can cause se­

vere arcing problems, which is unacceptable for sensitive substrates or polished samples.

One example is the high voltage electrode used for the Free Electron Laser electron injec­

tion gun at the Thomas Jefferson National Accelerator Facility. The electrodes are made

of stainless steel with special surface profile [2], polished first to 9 pm then to 1 pm to ef­

fectively suppress field emission from small imperfections or protrusions from the surface.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 Surface Processing by Plasma Source Immersion Ion Implantation 160

Beam Raster Target Manipulator Ion Extractor

Ion Source

5 Vacuum System

Figure 7.1. Traditional ion implantation setup.

HV Pulse Forming Network

Vacuum Pumps

Figure 7.2. Concept of PSII

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 Surface Processing by Plasma Source Immersion Ion Implantation 161

When such electrode is nitrogen ion implanted in the PSII system, extra attention has to be

paid to maintain the implantation process arc-free.

Nitrogen ion implantation has been found to improve steel surface hardness, corro­

sion resistance and wear resistance [117,118], producing higher nitrogen concentration and

wider hardened layer than low-voltage plasma nitriding [117]. Previous work in our group

established that nitrogen ion implantation suppressed held emission from small stainless

steel test samples [2]. Nitrogen ion implantation can also be used to produce TiN layers,

which are very hard and have superior wear resistance [119]. TiN also acts as diffusion

barrier between Cu and Si for microelectronics applications. Nitrides of other transition

metals such as Ta, Nb, and W are also attractive because their high thermal stability and

chemical inertness with Cu [120].

RFI PSII has the advantages of large-area, uniform, high-purity, long processing time,

and 3-D ion implantation (see Chapter 3 on large area RFI plasma source design for large-

scale PSII). Hence, it was used to perform nitrogen ion implantation into a variety of metal

substrates.

Low temperature, high dose nitrogen ion implantation into a selected group of high

purity transition metal substrates, including Ti, Ta, Nb, Mo, and SS, was conducted us­

ing the large-scale RFI PSII system. Surface morphology of implanted and bare substrates

were studied by optical microscope. Auger electron spectroscopy (AES) was used to study

the retained dose, concentration depth profile, and chemical composition of the implanted

samples. Surface hardness was also analyzed for different loads to study the surface hard­

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.2 Experiment 162

ening effect of nitrogen implantation. Highly polished stainless steel test electrodes were

also implanted for testing in the TJNAF FEL facility.

7.2 Experiment

Detailed information of our RFI source is given in Chapter 3. Here we provide a simple

description of the RFI PSII system to discuss the works of our implantation studies. The

Large-scale PSII system with planar RFI plasma source at ARC is shown in Fig. 7.3. It con­

sists three main subsystems: the implantation chamber, the planar RFI plasma generating

system, and the pulse-forming network (PFN).

The cylindrical implantation chamber is 0.61 m I. D., 0.51 m tall, with a 0.57 m in

diameter and 0.0254 m thick quartz window sitting on top of the chamber to work as the

dielectric media for RF coupling. Over the implantation chamber, a 0.4 m tall aluminum

cover shields the rest of the lab from RF radiation and protects the quartz window. It also

serves as the support structure for the RF matching network, water-cooling lines and the

0.43 m in diameter RF antenna that is in close contact with the quartz window surface.

The chamber is pumped by a Leybold-Heraeus Turbovac 450 turbo molecular pump at the

bottom of the chamber and by a CIT-Alcatel (Type2033) mechanical pump attached to a

side port. A Leybold D8A mechanical pump backs the turbo pump.

The pulse-forming network (PFN) is capable of delivering high voltage pulses of up

to 100 KV, 30 A at a repetition rate of 200 Hz. The high voltage outputs are connected to

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.2 Experiment 163

Quartz Insulating Plate

Power Supply

Faraday Shield Cooling Water

Permanent Magnets

HV Pulse Feedthrough Forming Network Shielding x

Mechanical Pump Mechanical Pump B Turbo Pump A

Figure 7.3. CWM large-scale RFI PSII system.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.2 Experiment 164

the sample stage inside the implantation chamber through a shielded HV feedthrough on

the side of the implantation chamber.

Metal samples to be implanted are placed on a 6 in ID stainless steel supporting plate

placed on the sample stage centered in the chamber. In this study, we used substrates of high

purity Ti, Ta, Mo, and Nb sheets, 25 mm x25 mm. Typical analysis by the manufacturer

indicated nominal purity of 99.6% for Ti sheets, 99.9% for Ta sheets, 99.9% for Nb sheets,

99.9% for Mo sheets. The Ti substrate is 1 mm thick. The other sheets are 0.5 mm thick.

Type 304 SS sheets with 0.03pm (Rmax) mirror finish were also implanted. Each of the

samples were half masked to provide areas that were not treated by nitrogen plasma for

comparison purposes.

Research grade nitrogen gas (99.999%) was introduced into the chamber to maintain

a pressure of 2.0 x 10-3 torr. The rf power used was 850 W- The samples were pulsed at

30 KV at 100 Hz. The samples were not actively heated. Previous investigation of the ion

plasma density distribution allowed us to place the samples in the area where the plasma

density was very uniform (see Fig. 3.9 for radial plasma density distribution). Thus the

implantation condition may be considered identical for each of the metal substrates. The

implantation time was 3 h-for a nominal dose of 1.75 x 1018/cm2.

The implanted samples were studied by AES depth profiling. The AES spectra were

investigated with an electron beam of 5 KeV and 1.0 ^ A. Surveys were acquired before

sputtering to determine elemental surface composition. Sputter depth profiling was per­

formed while monitoring all minor elements in the survey scan, along with the substrate

elements. Silicon and other chamber elements were also included in the list for monitoring

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 165

possible contaminants from the implantation system. A 3 KV, 30 fiA Ar+ ion beam was

used for sputter depth profiling. The beam was rastered across a 2 mm x2mm area, with a

incident angle of 55° to the sample normal.

The surface hardness was also tested with Shimadzu microhardness tester and was

compared to unimplanted samples. For each data point, ten indentations were made. The

average and standard error were calculated.

Optical microscopy was also used to compare the surface morphology of implanted

and unimplanted substrates.

7.3 Results and Discussion

7.3.1 Target Temperature

Target temperature is an important parameter for PSII applications. Higher temperature

usually results in more diffusion of the implanted species and in a deeper range. For masked

implantations, higher target temperature results in more lateral diffusion. In order to reduce

implantation time and reach higher dose, high power is usually applied, which can produce

high target temperature. High target temperature may or may not be beneficial for specific

applications. For instance, higher temperatures are desirable when diffusion of the species

is desired. On the other hand, high temperature is not desirable for plastic, rubber, and poly­

mer materials, or materials that with special microscopic structure which will be changed

by high temperature, such as alloys. Thus target temperature prediction or estimation is

important for applications design.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 166

It is difficult to measure the target temperature in a PSII apparatus. High pulsed bias

on the target is hard for thermocouples to handle. Optical measurement is limited by the

chamber and port setup.

The power is applied to the target by way of a series of high voltage pulses during

implantation. The PFN for our PSII system produces pulse of 0 — 100 KV, at 0 — 30 A,

at a repetition rate of up to 200 Hz. Ideally, each high voltage and high current pulse is

20 fis wide, with a flat top. The pulses provide high power density during the 20 /is period.

Between the pulses, the target loses heat by radiation to the chamber environment and from

conduction to the supporting stage. The net power input for individual pulse is quite small,

thus only a small temperature increase is expected for one pulse.

To calculate the equilibrium temperature or temperature rise over a longer period of

time, we estimated the time averaged power from four major sources: heating by high

voltage pulses, heat loss from conduction, radiation loss, and heat convection from the

plasma. Because the plasma is on before pulsing starts, it is assumed that the target has

reached an equilibrium temperature from plasma heating before the implantation begins.

This equilibrium temperature is taken as the initial temperature to calculate the temperature

rise from pulse heating. After the pulsing starts and the target is heated up, convection from

the plasma can actually cool the target and cause a temperature gradient. However, in these

studies, for the small metal sheets tightly coupled to the 5.970 in ID, 0.780 in thick stainless

steel holder, convective cooling is small and the thermal gradient may be considered to be

negligible [121]. Furthermore, it is assumed that for reasonable contact resistances and

surface areas, reasonable emissivity difference between the base plate and metal sheets, the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 167

temperature difference between the plate and the sheets is quite small, so the supporting

plate and samples assembly can be treated as a single body, simplifying the calculation.

The stage is connected to the HV input through a 1/4 in stainless steel tubing, inserted into

the side of the stage so heat conduction through the HV connection is negligible. However,

the target assembly is supported by a 2 in. ID, 3 in. long quartz rod, which is inserted

in a stainless steel fixture centered at the bottom of the chamber. The heat loss from the

conduction of the rod should be considered.

The lumped model is then justified for estimating the temperature for the PSII target

[121, 122]:

m c p ^ = Aq - Ata{T* - 2?) - KqAq(T - Tb)/Lq (Eq. 7.1)

where A is the surface area of the target, q is the time-averaged heating effect from

the pulsing on unit surface area, e is the emissivity of the radiating surface, a is the Stefan-

Boltzmann constant, To is the chamber temperature. The second term on the right hand

side represents the radiation loss from the target assembly. The third term represents the

heat conduction loss from the quartz rod, where Kq is the thermal conductivity of quartz,

Aq the contact area, Lq is the length of the quartz rod. One end of the rod is assumed to

have the same temperature as the target assembly, the other end have the same temperature

as the chamber bottom, which isTb. The heat loss due to the radiation from the rod surface

is ignored, since the value is small compared to the other terms on the right side of Eq. 7.1.

This assumption has also been justified afterwards using the obtained equilibrium target

temperatures.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 168

The initial temperature of the inside of the chamber and the target is assumed to be

400 K. The following material parameters are used [29]:

Material Density p Cp(J/kgK) Thermal Conductivity k (kg/m3) (at constant pressure at 25 °C) (W/mK) 304 SS 8020 460 30 Ti 4500 523 21.9 Ta 16400 140 57.5 Nb 8570 265 53.7 Mo 10200 251 138 Si 2328 705 124 Quartz 2200 752 1.38

In this model, the heating from the energetic ions is treated as a surface heat flux. A

volume heating process is only slightly more accurate since the ions actually deposit their

energy over a finite depth, which can be characterized by the typical heat conduction length

(considering transient heat conduction in a semi-infinite slab) [121]:

/ = 2x/«t

where I is the length over which significant temperature change has occurred, k =

k/(pCp) is thermal difiusivity, t is the pulse width. For the pulse width used in our implan­

tation, t = 20 ps, considering different materials properties, we have:

304 SS Titanium Tantalum Niobium Molybdenum Silicon I (pm) 25.5 27.3 44.8 43.5 65.7 77.7

Typical implantation depth for the above calculated materials is several thousand

angstroms, much less than the typical heat conduction length. Thus the volumetric heating

effects can be ignored and surface heat flux can be used in our calculation.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 169

70Q-

500-

400 0 2000 4000 6000 6000 10000 1800014000

Figure 7.4. Target temperature (in Kelvin) as a function of time in seconds. The applied pulse is 30 KV at 100 Hz. The target surface emissivity is 0.2.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 170

Eq. 7.1 was solved numerically using Maple. The typical results for stainless steel

plate with a pulse repetition rate of 100 Hz and different emissivities are shown in Fig. 7.4

(e = 0.2) and Fig. 7.5 (e = 1.0). In the plots, t represents elapsed time in seconds, T rep­

resents target temperature in Kelvins. For all the calculations, the input pulse is assumed to

be 30 KV, with an average current of 15 A and duration of 20^s. The secondary electron

emission coefficient at 30 KV was also considered to adjust the power input from pulsing.

As shown in Fig. 7.4 and Fig. 7.5, the target temperature rises sharply at the beginning,

reaches equilibrium value at a later time, depending on the balance of the heating, conduc­

tion heat loss, and radiation cooling process. It takes longer for target with lower emissivity

to reach equilibrium temperature. Lower emissivity also leads to higher ultimate temper­

ature. Fig. 7.6 shows the equilibrium temperatures obtained for different pulse repetition

rate and target emissivity. Higher repetition rate of pulse provides more time-averaged

heating of the target, which leads to higher equilibrium temperature especially with target

having low emissivity. However, at pulse repetition rate of 50 Hz, the equilibrium temper­

ature is only around 259°C if the surface emissivity is 0.6, which is typical for polished

steel; or 236°C if the emissivity is 0.8, for steel with a shiny layer of oxide [29]. Fig. 7.6

also applies to all the other materials listed above, since the ultimate temperature is a func­

tion of emissivity, chamber temperature, and the input heating power. Target equilibrium

temperature is not a function of target mass and heat capacity.

Lower chamber temperature results in slightly lower equilibrium target temperature,

due to more radiation heat loss to the environment, as shown in Fig. 7.7.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 171

560- 540- 520-

440- 420- 400 0 2000 4000 6000 B00Q 10000 14000 18000

Figure 7.5. Target temperature (in Kelvin) as a function of time in seconds. The applied pulse is 30 KV at 100 Hz. The target surface emissivity is 1.0.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 172

1000

* 900 - 50 Hz £ O 100HZ 3 - w - 200 Hz 03 a> 300 - a . E a) l- a3 700 - E> 0) I- | 600 ■

£ 50°-

400 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Emissivity

Figure 7.6. Target equilibrium temperature for different pulse repetition rate and surface emissivity. The chamber temperature used is 400 K.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 173

1000

0) 0 5 700 ■ 01w I- E ~ 600 ■ n '5

^ 500 -

400 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Emissivity

Figure 7.7. Target equilibrium temperature for different pulse repetition rate and surface emissivity. The chamber temperature used is 300 K.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 174

The use of time-averaged heating by HV pulses for our calculations is justified by

the following considerations. The temperature rise from one 20 fxs wide pulse is calcu­

lated to be 0.18 K. Once the pulsing starts, the target temperature ratchets up, increasing

during the pulse-on period, and decreasing during the pulse-off period. However, because

the radiation heat loss is small compared to heating at low target temperatures, the target

temperature increases for each pulse cycle. At high target temperatures, the radiation loss,

which becomes the dominant heat loss mechanism, increases to be comparable to or equal

to the heating from one HV pulse. Therefore the net temperature change for each pulse

cycle decreases and the target temperature reaches a quasisteady state in which the temper­

ature at the end of the cycle is identical to the temperature at the begining of the cycle. The

target temperature history is illustrated in Fig. 7.8, where the time period and heating effect

are exaggerated to show this effect. For the actual PSII conditions we are interested in, the

heating period is much shorter (20 jis) compared to the cooling period, which is 4980 fis

for a pulse repetition rate o f200 Hz, allowing more time for the target to cool. The temper­

ature swing is also much smaller for each pulse (0.18 K). The actual temperature history

closely resembles a smooth curve.

It is concluded from the above calculations that active target cooling is necessary

for higher pulse repetition rate and for heat sensitive samples. One way of doing this is

to use a water-cooled sample stage. If the stage is kept near room temperature by active

water-cooling and is in good thermal contact with the back of the sample assembly, the

cooling effect can be estimated. Here, the same target assembly is considered as for above

calculations. Assume that the whole bottom of the assembly is cooled by the sample stage.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 175

440

430 -

* 420 ■

« 410

400 ■

390 0 20 40 60 80 100 120 140 T (seconds)

916

914 •

912 •

- 910 • £ -(0 908 • 01 §■ 906 • 4) H 904-1 O)4) (5 l- 902

900 -

896 0 40 6020 80 100 120 140 T (seconds)

Figure 7.8. Schematic of temperature history for the target. Time periods and temperature swings are exaggerated relative to typical PSE1 target temperature histories.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 176

Only the top and side of the assembly are implanted, heated, and cooled by radiation heat

loss, so a cooling term is added to the equation:

j r p me,,— = Aq — Ae

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 177

550

500

T(t)450

400

350

300- 0.2 0.4 0.6 0.0

Figure 7.9. Target temperature history as a function of time. Assume that the target temperature has reached 600 K by pulse heating before cooling is turned on at t=0. Assume that 0.1 mm thick silver layer is used for heat conduction between the target and the cooled stage, which is at 298 K. by active water cooling.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. r.3 Results and Discussion 178

600

550-

500-

400

350

300- 0.2 0.4 0.6 0.8

Figure 7.10. Target temperature history as a function of time. Assume that the target temperature has reached 600 K. by pulse heating before cooling is turned on at t=0. Assume that 0.1 mm thick indium layer is used for heat conduction between the target and the cooled stage, which is at 298 K by active water cooling.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 179

Ac is the contact area between the target assembly and the cooling stage, Ts is the

cooling stage temperature, and h is the heat transfer coefficient of the heat conducting

material used between the stage and the assembly. Assume that there is a 0.1 mm thick

heat conducting material between the target and the stage, and the stage surface is kept

at 298K by water cooling. Also assume the chamber wall temperature to be 400 K. and

the target has reached 600 K. because of the pulse heating. If the heat conducting material

is silver, the heat transfer coefficient would be 4.3 x 106 W /rrrK, obtained by dividing

thermal conductivity by the layer thickness. Calculation gives the temperature history after

the cooling is turned on in Fig. 7.9, showing that the target temperature would drop to the

stage temperature (298 K) in less than a second (around 0.1 second). If Indium is used, the

heat transfer coefficient would be 8.16 x 10s W /m 2K for a 0.1 mm thick layer. Similar

calculations can be performed and the result is shown in Fig. 7.10. The target assembly

can also be cooled to 298 K. in less than a second (around 0.5 second). Since copper or

other good heat conducting materials are usually used to construct cooling stages, the time

it takes for the stage to reach 298 K was ignored. Similar calculations were also performed

for targets that had already reached higher temperatures (900 K or 1000 K) by pulse heating,

showing that it takes less than a second for the target to be cooled to 298 K.

A calculation considering the heating effect of the individual HV pulse was per­

formed since the cooling stage was found to cool the target to stage temperature in very

short period of time. It was assumed that the starting temperature for the stage was 400 K.

Eleven HV pulses were applied to the target raising the target temperature to 401.98 K, be­

fore the cooling was turned on. The pulse repetition rate was 200 Hz. The pulse-on time

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 180

was 20 /is and the pulse-off time was 4980 fis for each heating cycle. Cooling was constant

once it was turned on. The temperature history (Fig. 7.11) showed that the target temper­

ature dropped from 401.98 K. to 298 K. in about 0.1 second, confirming the results above

obtained using time-averaged heating of the HV pulses (Fig. 7.9). Enlarged plots of the

temperature curve were also made and are shown in Fig. 7.12 and Fig. 7.13. As shown

in Fig. 7.12, the target temperature rise was 0.18 K for each pulse. Because the radiation

loss was small at low temperatures (around 400 K), temperature drop during the pulse-off

period was not significant. After the cooling was turned on (as shown in Fig. 7.13), heat

generated by each pulse was slightly smaller than the conduction heat loss to the cooled

stage, resulting in small temperature drop (< 0.1 K) for the pulse-on period. The temper­

ature drop for the pulse-off period started from 26.36 K at t = 6 x 10-2 s, decreased to

0.14 K at t = 1.5 x 10- l s, as the temperature difference between the target and the stage

decreased.

7.3.2 Profile Calculation of Nitrogen Concentration

The profile calculation of low dose implantation helps to understand how implanted ions

of the same energy interact with different substrates. Nitrogen concentrations in Ti, Nb,

Mo and Ta substrates were calculated by Profile Code. Profile Code was developed by Im­

plant Sciences Corporation, for rapid prediction of implantation profiles under a variety of

conditions, such as different substrate materials, implantation species, doses, incident an­

gles, and implantation voltages. The Profile Code results are usually equivalent to those

generated by Monte Carlo programs, such as TRIM™. The calculations of Profile Code

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 181

420

400

a) 3 ra 360 -

£ 340 - 4b* aia) ra \- 320 -

300 ■

280 0.0 2.0e+4 4.0e+4 6.0e+4 8.0e+4 1.0e+5 1.2e+5 1.4e+5 1.6e+5

Time (microseconds)

Figure 7.11. Calculated target temperature history showing the heating from a train of pulses. The target starting temperature was assumed to be 400 K. Eleven HV pulses were applied before the stage cooling was turned on (0 - 5.5 x 104us). The pulse repetition rate was 200 Hz. Cooling was turned on at 5.5 x 104us when the target was at 401.98 K.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 182

0.20 8 6

14

12

10

0.08

0.06

0.04

0.02

0.00 0 20 40 60 80 100 Time (microsecond)

2

0 -i------1------1------1------1------1------0 10000 20000 30000 40000 60000 60000 Time (microsecond)

Figure 7.12. Target temperature history calculated showing the heating effect of individ­ ual HV pulses. The pulse repetition rate was 200 Hz.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 183

388

386

384

382

380

378

376

374

372 S9000 59920 59940 59960 59980 60000 60020 60040 Time (microsecond)

380

370

* 360 83 2 350 & E £ 340 OJV 5 330

320

310 55000 60000 65000 70000 75000 80000 85000 Time (microsecond)

Figure 7.13. Target temperature history which considers the heating effect of individual HV pulse and the cooling from the stage. The pulse was 20 [is wide and the repetition rate was 200 Hz. The target temperature was 401.98 K after heated by 11 HV pulses. The cooling was turned on at T = 55000 [is.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 184

are usually completed in a matter of seconds instead of hours as required by other codes

[102]. It is considered a powerful research tool for experiment design, and for understand­

ing experimental depth profiles. For the following calculations, an implantation voltage

of 30 KV and applied dose of 1 x 1017/cm 2 was used. As shown in Fig. 7.14, nitrogen

penetrates deepest into the Ti substrate for the same applied dose and implantation volt­

age. Its peak concentration is 19.1% at 584 A . Retained dose is 92.4% of the applied dose,

or 9.24 x 1016/cm 2. The nitrogen peak concentration is 24.1% at 304A for Nb substrate,

19.3% at 287A for Mo, and 20.4% at 220 A. It is obvious that the peak position is closer

to the surface for substrate with larger z values. The sputter loss is 61 A for Ti, 78 A for

Nb, 66A for Mo, and 62 A for Ta. Retained dose is 83.5% for Nb, and 74.5% for Mo,

and 64.9%for Ta, which is higher for substrate with low-z elements. Ions are assumed to

be incident perpendicularly on the substrate surface. These results are valid for pure sub­

strates, without any surface oxide or carbide layers. With the existence of these surface

layers, the sputtering coefficient would be significantly lower than the prediction of Profile

Code [102],

7.3.3 Characterization of Implanted Samples

Optical Microscopy Study of the Sample Surface

The implanted Ti sheets show a golden color which is characteristic of stoichiometric

TiN. Optical microscope pictures of implanted and un-implanted Ti substrate areas are

shown in Fig. 7.15 and Fig. 7.16. They indicate that the implantation process helped to

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced usrts h mlnainvlaei 0K n h mlnainds s1 101T/cm 2. implantation voltageisKV The 30 x 1 and theimplantationissubstrates. dose Figure 7.14. Profile Code calculation of nitrogen concentration in Ti, Nb, Mo and Ta and Mo Nb, Ti, in concentration nitrogen of calculation Code Profile 7.14. Figure

Nitrogen Concentration (%) 30 20 25 15 10 0 5 0 Mosubstrate Ta substrate Ta Tisubstrate Nb substrate Nb 200 0 600 400 . Results and Discussion 7.3 Depth(angstroms) 800 1000 1200 1400 185 7.3 Results and Discussion 186

smooth out the sharp features and possibly to anneal the surface. This is important for the

implantation treatment of FEL electrodes for suppressing field emission. Ti is also being

used to make the FEL electrode because of its high hardness. The implanted nitrogen

forms nitride and hardens the Ti surface. The conditioning and smoothing effect of the

PSII process also removes sharp surface protrusions, which are believed to be locations of

high field strength and sources of field emission.

The nitrogen implanted Ta shows a dark blue color. Optical microscope pictures

do not show apparent differences of surface morphology between the implanted and un­

implanted area (Fig. 7.17 and Fig. 7.18).

Nitrogen implanted Mo shows a silver metal color. Compared to the implanted area,

the un-treated area shows a yellow-green hue. Optical microscope pictures of implanted

and un-implanted Mo surface are shown in Fig. 7.19 and Fig. 7.20. The implanted

area is apparently smoother than the un-implanted area. The implantation process possibly

annealed the Mo surface, too.

The nitrogen implanted Nb surface shows a pale golden color. However, the optical

microscope study did not show any apparent difference for the implanted and un-implanted

substrate (Fig. 7.21 and Fig. 7.22).

Nitrogen implanted stainless steel shows a uniform dark brown color. Optical micro­

scope picture of the implanted and un-processed sample surfaces are shown in Fig. 7.23

and Fig. 7.24. There are no signs of arcing or other damage on the implanted area. The

implanted substrate has a surface smoothness at least as good as the surface finish of the

un-treated substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 187

Figure 7.15. Optical microscope picture of nitrogen implanted Ti substrate.

Figure 7.16. Optical microscope picture of un-implanted Ti substrate

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7J Results and Discussion 188

Figure 7.17. Optical microscope picture of un-implanted Ta substrate.

Figure 7.18. Optical microscope picture of nitrogen implanted Ta substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 189

Figure 7.19. Optical microscope picture of nitrogen implanted Mo substrate.

Figure 7.20. Optical microscope picture of un-implanted Mo substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 190

Figure 7.21. Optical microscope picture of un-implanted Nb substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 191

Figure 7.23. Optical microscope picture of 0.03 pm minor finish 304 stainless steel suh strate.

Figure 7.24. Optical microscope picture of nitrogen implanted mirror finish stainless steel substate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 192

Surface Hardness Study

The surface hardness of the implanted and un-implanted substrates are studied using

different load values. Because the implanted layers are very thin compared to the indenta­

tion depth, the Vickers criteria were not rigorously satisfied. Some attempt to compensate

for this can be made by measuring the hardness value as a function of load (indentation

depth) and extrapolating the resulting curves to their values at zero loads. The general

problem of measuring the hardness of ultra-thin, very hard layers over soft substrate has

been discussed in reference [61]. The measurements reported here can be used as a qual­

itative indication of the implanted layer hardness, which show that the implanted samples

are indeed harder (probably much harder) than the bare samples (Fig. 7.25, Fig. 7.26, Fig.

7.27, Fig. 7.28, and Fig. 7.29) The surface hardness results for Mo substrate are shown in

Fig. 7.27. Because of the surface roughness, the reading at I5g load is very unreliable and

thus is not included in the graph. TiN layer that has larger ratio of thickness to indentation

depth yields larger hardness values, confirming this qualitative result (see below for AES

depth profiles of implanted nitrogen concentration).

AES Depth Profile of Implanted Nitrogen

The AES spectrum of the implanted Ti surface is shown in Fig. 7.30. Elements

present on the surface in trace amounts include carbon, nitrogen, oxygen, titanium, cal­

cium, sodium, aluminum, copper, and sulfur. Because of surface absorbed ambient gas

species, carbon, nitrogen, and oxygen have prominent peaks. The AES depth profile of im­

planted nitrogen is shown in Fig. 7.31. The Ti-N peak overlap and the influence of oxygen

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 193

— Implanted Area —O— Un-implanted Area 3600- 3400- 3200- £ 3000- 2800- 2600- p 2400-

2200 -

2000 - 1800- 1600- 1400 -

0 20 40 60 80 100 Load (g)

Figure 7.25. Surface hardness test results for nitrogen implanted and un-implanted Ti substrates.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 194

— Implanted Area 2000 -, —O— Un-implanted Area 1900- 1800- 1700- Oj 1600 - (0 1500- (A Q) 1400- ■oC (uk._ 1300 - X(/) 1200- ® 1100 -

> 1000- 900- 800-

0 20 40 60 80 100 Load (g)

Figure 7.26. Surface hardness test results for nitrogen implanted and un-implanted Ta substrates.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 195

— Implanted Area 4300-, —O— Un-implanted Area 4200- 4100- ^ 4000 - 2 3900- 3800 - in in 0) 3700- c ■O_ 3600- coL_ X 3500- -W 3400- a) 3300- > 3200- 3100- 3000 20 30 40 50 60 70 80 90 100 110 Load (g)

Figure 7.27. Surface hardness test results for nitrogen implanted and un-implanted Mo substrates.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 196

— Implanted Area —O— Un-implanted Area 1600-

1500-

flj 1400-

1300 -

1200 - 7 3

1100 -

1000 -

900-

800-

0 20 40 60 80 100 Load (g)

Figure 7.28. Surface hardness test results for nitrogen implanted and un-implanted Nb substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 197

— Implanted Area 2350- —O— Un-implanted Area 2300-

2250 (0 Q- 2200

OT 2150 V) c 2100- •E n 2050- « 2000

1900

1850 - 1- —I— —r~ “T" 20 40 60 80 100 Load (g)

Figure 7.29. Surface hardness test results for nitrogen implanted and un-implanted mirror finish 304 stainless steel substrate.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 198

on the T im /T im peak ratio was considered when calculating the nitrogen concentration

[101]. Nitrogen concentration is around 60-64% percent for the top surface (0 — 950A),

except for the topmost 2 0 0 A range, where the higher concentration of oxygen from the

surface oxides prevent more nitrogen from reacting with the titanium. The oxygen concen­

tration decreases from about 25% to below 3% in 0 to 800A range. Nitrogen concentration

slowly decreases after a depth of 1 0 0 0 A . Carbon, aluminum, and silicon are of very low

concentration throughout the sputtered range; they are intrinsic impurities in the metal. Sul­

fur was not plotted for the depth profile because it was not detected except at the surface,

where it was only 0.5%.

The almost constant concentration of implanted species for the top depth range is

characteristic of high dose implantations. The concentration of implanted species usually

saturates at the stoichiometric level for each depth. This can also be illustrated by cal­

culation results from Profile Code, as shown in Fig. 7.32 and Fig. 7.33. At a dose of

6 x 10l7/cm 2, the nitrogen concentration begins to saturate for depth around 800 A . As the

dose is further increased to 1 x 10l8/cm2, the small gap at the top surface is further filled

and the layer thickness increased from around 1500 A (Fig. 7.32) to 2000 A (Fig. 7.33).

The implanted layer has a constant nitrogen concentration for most of the implanted range,

which is very similar to a deposited stoichiometric TIN film.

It i s noticed that the nitrogen concentration in the Ti substrate for the top 2 0 0 A to

1 0 0 0 A range is higher than 50%, the stoichiometry level of TiN , and therefore higher

than the45% level predicted by Profile Code. Titanium is often used as vacuum pumping

material [123,124] wherein evaporative or sputtering processes are used to produce nascent

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 199

1 500 N-im planted Ti as received 1 ooo

500

3

• 1000

•1 500

•2000

200 400 600 800 1000 1200 1400 1600 1800 2000 Kinetic Energy («V)

Figure 7.30. AES spectrum of the implanted Ti surface.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 200

1.0-1

0.9-

0.8- Nitrogen c 0-7- o Titanium 0.6- (Q Calcium g 0.5- Silicon Aluminum

o 0.3- Oxygen £ Carbon O 0.2- < 0.1 -

0.0 -

- 0.1 0 500 1000 1500 2000 Depth (Angstroms)

Figure 7.31. Concentration depth profile of a nitrogen implanted Ti substrate obtained by AES depth profiling

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 201

Ti films, which provide gas-absorbing power. We believe that solid solutions of nitrogen

are also formed on our Ti substrates. The micro-structure of the substrate also influences

the retained concentration of nitrogen. It was found in the sample preparation process

that the Ti sheet is porous and mechanically weak. It is also quite possible that excess

nitrogen can be trapped between the Ti particles and the sputtered particle surfaces, causing

higher nitrogen concentration in these samples. Other studies [120] have found nitrogen

concentrations as high as 70% in ion implanted and high temperature annealed TiN layers.

Our implanted nitrogen concentration profile has an elongated tail. This may be caused by

the elevated temperature. As shown in Fig. 7.6, at a repetition rate of 100 Hz and emissivity

of 0.8, the target temperature can reach ~300°C, sufficient to allow for enhanced diffusion

to occur.

The AES spectrum of nitrogen implanted Ta is shown in Fig. 7.34. Tantalum,

nitrogen, carbon, oxygen, and calcium are detected. Carbon and oxygen were probably

mainly from the surface absorbed gas molecules, resulting from long duration exposure to

air prior to analysis. The AES concentration depth profile are shown in Fig. 7.35. The

nitrogen concentration is around 40% for the top 450A range, decreasing slowly to below

10% after 1150A. The topmost 50A range also contains less nitrogen due to the higher

oxygen concentration in that range. It also shows a long tail, probably due to diffusion

caused by elevated substrate temperature.

The AES spectrum of an implanted Mo sample is shown in Fig. 7.36. It includes

peaks for Mo, C, N, Cu and Al. Again, the high concentration of C and O on the surface

can be explained by surface absorbants. The AES concentration depth profile is shown in

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 202

70

60

50

40

30 O) 20

10

0 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Depth (Angstroms)

Figure 7.32. Profile Code calculation of nitrogen concentration in a Titanium substrate for implantation voltage of 30 KV and dose of 6 x 10 17/cm 2.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 203

60 -

50 - C o (0 Z 40 i § O 30 ■ C

0 500 1000 1500 2000 2500 Depth (Angstroms)

Figure 7.33. Profile Code calculation of nitrogen concentration in a Titanium substrate for implantation voltage of 30 KV and dose of 1 x 1018/cm 2.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

c/s -1500 -1000 1500 1000 -500 500 Figure 7.34. AES spectrum of AESspectrum7.34.Figure a of implanted nitrogen Ta substratesurface. 0 40 0 80 00 20 40 60 80 2000 1800 1600 1400 1200 1000 800 600 400 200 o i t . Resvlls Discussion and 7.3 iei Enry (eV) nergy E Kinetic a 2.9 IT. 3 Cal 19.4 01 .a 9 2 Cl Ml 32.5 7*2 *. Atonic s received as Ta planted N-im

• .1* ' 1 ' . 204 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Atomic Concentration (%) 100 O ■SO SO 90 0 -• 40 50 70 20 1 30 o - - 20 0 60 0 10 10 10 10 10 2000 1800 1600 1400 1200 1000 800 600 400 200 0 I Figure 7.35. AES depth profile of FigureprofileAES 7.35.depth nitrogen implanted a of Ta substrate. \01 ' Si2' lal 7 3 Results and Discussion .3 pte et (AjDepth Sputter 0 sputter rate a r r e t t u p s a T 2 i0 d S te n la p N-im 205 7.3 Results and Discussion 206

Fig. 7.37. The nitrogen concentration decreases monotonically from 17% at the surface

to about 3% at 200A depth. Compared to the Nb implanted at the same conditions, the

nitrogen concentration is low and the range is confined to the topmost layer. A possible

reason for the very low nitrogen concentration is the existence of thin but dense carbide

and oxide layer on the Mo surface. The carbon concentration is more than 20% for the top

50 A range and the oxygen concentration is over 20% from 10A to llOA range. Compared

to Mo,the surface oxygen and carbon concentration are lower for Nb substrate, which has

a much higher and wider distribution of nitrogen (see below).

AES spectrum of the implanted Nb surface is shown in Fig. 7.38. The elements

shown are nitrogen, carbon, oxygen, and niobium. The concentration obtained by AES

depth profile is shown in Fig. 7.39. The top 300A layer shows a nitrogen concentration

ofaround 50%, which decreases slowly to about 1 0 % at 1400A. Nitrogen concentration is

lowerthan 50% at the top surface where the oxygen and carbon concentration are higher.

Throughout the sputtered depth, except for the top 1 0 0 A range, there is an almost constant

concentration of oxygen (around 10%), carbon (around 6%) and silicon (around 4%).

The AES spectrum of the implanted SS surface is shown in Fig. 7.40. Elements

detected are oxygen, ion, carbon, nitrogen, silicon, aluminum, calcium, and sulfur. AES

depth profile (Fig. 7.41) shows that the nitrogen concentration peaks at about 2 0 0 A , with a

peak value of 19%. The nitrogen concentration stays above the 10% level until 1800 A , then

slowly decreases to below 3% around 3000 A . The condition through the top 0-600A range

is more complicated. The Ni peak concentration (22%) is found at 400A depth while Cr

peaks (27%) at 2 0 0 A . For depths larger than 600A, the Ni concentration fluctuates around

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 207

N-implanted Mo 1000 as received

500

0 Mol Cu Mo

-500

A t o n ic * Cl 01 Ml -1000 Mol Cul a i :

-1500

200 400 600 800 1000 1200 1400 1600 1800 2 0 0 0 Kinetic Energy (eV)

Figure 7.36. AES spectrum of a nitrogen implanted Mo substrate surface.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Atomic Concentration (%) 100 Figure 7.37. AESFigure7.37. profiledepth nitrogen of a implanted Mosubstrate. 90 40 60 70 80 20 30 50 0 200 . Results and Discussion 7.3 400 600 S putter Depth Depth putter SI0 2 sputter rate sputter 2 SI0 -mpatd Mo N-im planted 800 (A) 1000 208 1200 401600 1400 7.3 Results and Discussion 209

N-im planted Nb 1500 as received

1000

500

Nb 0 Nb Nb Nb

-500 Aeomic \ HI 34.

Cl -1000 Hbi

-1500

-2000 200 400 600 800 1000 1200 1400 1600 1800 2000 Kinetic Energy (eV)

Figure 7.38. AES spectrum of a nitrogen implanted Nb substrate surface.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Atomic Concentration (%) 0 0 1 g0 SiO 2 sputter rate sputter 2 SiO g0 70• 80 40 - 40 60 0- N' 50- 20 30■ 0- c _ — _ ct • 10- o Figure 7.39. AESdepth profile7.39.Figure of implantednitrogensubstrate.a Nb 20 0 60 0 10 10 10 10 1800 1600 1400 1200 1000 600 600 400 200 0 i ot ~ o t ' ------I ------mpatd Nb N-implanted . Results and Discussion 7.3 pte et (A) Depth Sputter __ 210 N1 2000 7.3 Results and Discussion 211

8% and the Cr increases slowly to around 20% level. Oxygen concentration decreases

from 42% at the top to about 2% at 500A depth. Higher silicon concentration is also found

in the top 100A range, which may have been introduced from polishing and processing.

The nitrogen peak around 200 A is probably due to the combined effects of implantation

parameters, peaked Ni, and Cr concentrations for the 0-700 A range, and the high surface

oxide concentration.

Compared to the nitrogen implanted Ta and Nb, which were implanted with the same

experimental parameters, and having maximum nitrogen concentration of 42% and 50%

respectively, the nitrogen concentration of the stainless steel is much lower (<20%). How­

ever, the penetrated depth is deeper than that of Ta and Nb, which decreases to below 10%

at 1000 A and at 1500 A respectively. Except for the top 300 A layer, the nitrogen concen­

tration curve is much flatter, staying around the 10% level for a range of about 2000 A . To

obtain higher dose for stainless steel, a longer implantation time must be used.

Among the implanted samples, the largest hardness increase was found on the nitro­

gen implanted Ti substrates (Fig. 7.25), which was also the one having the highest nitrogen

concentration and the largest implanted layer thickness, according to AES depth profiling.

Under the same implantation conditions as used for Ti substrates, Mo and SS samples have

lower retained doses, which caused insignificant hardness increases compared to bare sub­

strates. The implanted layer shape is similar for nitrogen implanted Ta and Nb samples,

with Nb having slightly higher N concentration (Fig. 7.35 and Fig. 7.39). However, both

showed very little hardness increase compared to bare substrates.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 212

2500 N-implanted SS as received 2000

1500

1000

500

c/s

-500

-1000 A to m ic \ O t 3 5 .6

-1500 11.2 S12 A1Z -2000 C a l

-2500

200 400 600 800 1000 1200 1400 1600 1800 2000 Kinetic Energy (eV)

Figure 7.40. AES spectrum of a nitrogen implanted stainless steel substrate surface.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced

Figure 7.41. AES depth profile of profiledepth AES nitrogen a of substrate. steel 7.41. Figurestainless implanted Atomic Concentration (%) 100 ' ° - 0 4 60 90 50 70 80 30 20 20 0 — " 10 0 50 00 50 00 50 00 50 00 4500 4000 3500 3000 2500 2000 1500 1000 500 0 S2 - — -- - — - - - - — - - - - \Si2 - •' t- i I r _ Cr2 i Cr 2 Nit Fe2 7 3 Results and Discussion .3 __ ----- Sputter Sputter Depth(A) S i0 2 sputter rate sputter 2 i0 S SS -implanted N Nil - Fe2 213 7.3 Results and Discussion 214

AES study of the implanted Ti substrates confirmed that the RFI was operating in

the inductive mode, since there are no signs of Si ( > 2 film deposition on the Ti substrates.

The very low concentration (< 3%) of Si throughout the sputtered depth was believed to

be residual impurity in the Ti substrate, which was found at the same level and with similar

distribution in the AES depth profile for un-processed Ti substrate (Fig. 6.4).

It was concluded that the retained implantation dose varied depending on the type of

substrate material, its micro-structure, and the existence of surface carbide and oxide layers

under the same implantation conditions. Surface cleaning may be necessary to remove

surface layers for certain applications, so that desirable implantation depth profile can be

obtained with better prediction and control.

Nitrogen Implanted FEL Stainless Steel Electrode

Traditional plasma nitriding of stainless steel is usually done at elevated temperature

to facilitate nitrogen diffusion, to increase nitride layer thickness, and to decrease treating

time [118,125]. Stainless steels have excellent corrosion resistance due to the native pas­

sivation layer, resulting from chromium enhancement at the surface, but they do not have

high surface hardness. High dose nitridation is said to be effective in enhancing the SS sur­

face hardness. However, higher temperature causes the formation of CrN and a drop of

corrosion resistance of the stainless steel surface.

PSII is advantageous in perform nitriding of the stainless steel surface because it

can be operated below the critical temperature of 673K for CrN formation [118]. Low

temperature is also important for keeping the original target dimension and shape, and

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.3 Results and Discussion 15 2

retaining desirable bulk material microstructure. In addition to being a low-temperature

process, PSII is capable of 3-D uniform dosing, which is important for the surface treatment

of FEL stainless steel electrode for suppressing field emission.

Nitrogen implantation of large FEL electrodes may be useful in suppressing field

emission and in enhancing the performance of FEL electron injection gun. Preliminary

study showed PSII treatment on small cm scale polished stainless steel test samples reduced

field emission [2]. However, the results were obtained at narrow gaps and low voltages, so

more research, especially tests of large area implanted electrodes, is needed to study the ef­

fectiveness of nitrogen PSII. In addition, our previous PSII system used hot filament or DC

glow plasmas, which produced low ion plasma densities, large-spatial non-uniformities,

and were possible sources of contamination. These features are unsuitable for large-area

3-D FEL electrode implantation. RFI PSII makes high voltage breakdown tests of full-size

3-D electrodes possible. Highly polished 3-D electrodes have a surface finish of from 1 pm

to 9 pm. For polished large area FEL electrodes without nitrogen implantation, turn-on

voltages can vary by factors of 2 or more for similarly prepared electrodes. Anomalous

field emission from the electrodes is unpredictable in nature. PSII nitriding can help to in­

crease surface hardness, and to remove (condition) surface protrusions, which are the sites

of strong field emission. Higher surface hardness values may not be the reason for reduced

field emission.

A 1/3-scale FEL test electrode was successfully nitrogen implanted during these stud­

ies using the large-scale CWM RFI PSII. As shown in Fig. 7.42, the test electrode has spe­

cial 3-D structure and is highly polished. The implanted area shows very uniform color,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.4 Summary 216

without any signs of arcing or other kind of damage, indicating very uniform ion implanta­

tion.

The PSII system also offers other exciting possibilities for alternative 3-D treatments

of FEL electrodes. High quality SiC >2 films have been deposited on 0.15 m ID electrode

with Rogowski surface profile, with constant nitrogen concentration depth profile in the

films. This is discussed at length in Chapter 6. The PSII system makes it possible to

control the spatial distribution of nitrogen in the deposited films. Nitrogen sputtering of

the electrode surface can also be performed for surface oxide/carbide cleaning. Nitrogen

implantation at the surface can be performed prior to film deposition for interface adhesion

promotion. The large-scale RFI source can also be used to excite large uniform plasmas

of other gas species to perform chemical vapor deposition on large FEL electrodes. With

uniformly elevated target surface temperature of around 400°C, high quality DLC films can

also be deposited on FEL electrodes (see Chapter 2 on DLC deposition for applicability to

turbine engine applications). These prospects will be discussed in Chapter 8.

7.4 Summary

Nitrogen ion implantation has been successfully performed using the CWM large-scale RFI

PSII onto a variety of substrates, including Ti, Ta, Mo, Nb and stainless steel sheets, and

three dimensional highly polished stainless steel FEL test electrode. To help to understand

surface analysis results, target temperature rise due to implantation pulses was calculated

for different implantation repetition rate, target surface conditions, and stage setup. Implan-

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.4 Summary 217

Figure 7.42. Nitrogen implanted FEL test electrode sitting on a support flange. It is 2.S in. ID. The highly polished stainless steel surface shows uniform color. Its surface is also free of arcing or any other kind of damage.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7.4 Summary 218

tation depth profiles for different materials were calculated by Profile Code. The surface

morphology of the implanted and un-implanted metal sheets were studied by optical mi­

croscopy for comparison. These studies showed that nitrogen PSII can help to smooth the

Ti and Mo substrate surface. Surface hardness of implanted samples were studied using

different load values and compared to that of bare substrates. Nitrogen implanted Ti sub­

strate showed the largest hardness increases. Hardness increases were lower for other metal

surfaces. AES was used for implanted substrate elemental composition study and surface

depth profiling. AES depth profiling of the implanted and un-processed Ti substrates con­

firmed that there is no SiOo contamination from the PSII system. The nitrogen implanted

Ti substrate has the highest retained dose and highest maximum nitrogen concentration,

which also has the highest hardness increase compared to the bare substrate. Surface oxide

and carbide layers caused low nitrogen concentration and shallow implantation layer for

Mo substrate. Under the same implantation parameters, the nitrogen concentration for the

implanted layer varied because of the substrate structure, atomic number, and existence of

surface oxide and carbide layers.

Nitrogen ion implantation of 3-D FEL test electrode was also successfully performed.

The implanted electrode shows very uniform color, without any signs of arcing or other

damage. The ability to perform large-area, 3-D implantation makes it possible to test the

surface treatment effects on large-area, real size FEL electrodes. Correlated with surface

analysis results, this information can help the search for the best surface treatment recipe to

suppress field emission, and lead to an understanding of the mechanism of field emission

from these highly polished high voltage electrodes.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 8

Conclusions and Future Work

8.1 Conclusions

A planar RFI PECVD system was designed and built for large-area, uniform DLC films

deposition. Ti-6A1-4V substrates have been coated with DLC films without surface pre­

treatment and with Ti/MoS 2 films with only a simple surface pre-cleaning. Both kinds of

films exhibited much better tribological properties than the plasma sprayed Cu-Ni-In coat­

ing that is currently used to protect Ti-6A1-4V substrate against fretting and wear in turbine

engine applications. Optical investigation was found to be valuable for tribological coating

characterizations. Variable angle spectroscopic ellipsometry was used to measure optical

properties and thickness of the PECVD DLC films. The optical properties of DLC film

helped to quickly evaluate the film quality and to indicate whether further destructive and

time-consuming tests were warranted. These measurements were also correlated with inter­

ference color and visible appearance of the films to provide a fast assessment of thickness

and smoothness.

219

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 Conclusions and Future Work 220

A large-scale planar RFI plasma source was successfully designed and built, to fully

realize the large-area uniform plasma generating capability of RFI source and the potential

of PSII for large scale, 3-D, uniform surface processing. The design and materials problem

in the scale-up process have been successfully solved. The generated plasma was charac­

terized by Langmuir probe and OES in detail. The large-scale RFI source is very stable,

producing plasma density higher than that of the previous hot filament plasma or DC glow

discharge, with greatly improved spatial uniformity. The successful operation of large-

scale RFI makes 3-D, large size, uniform, contamination-free plasma source immersion

ion implantation possible.

The AFOSR MAGIC code was also for the first time used to model plasma source ion

implantation process. Taking advantage of the many successfully integrated PIC techniques

of MAGIC Code, the PSII modeling presented in this dissertation addressed different target

geometries and boundaries, materials, plasma parameters, illustrated sheath formation and

evolution, field distribution, ion and electron trajectories, ion incident angles, and dose

distributions, which are critical information for PSII experiment design and understanding

of PSII process. MAGIC modeling of PSII is advantageous compared to 1-D or pseudo

2-D analytical models, for its 3-D and multiple information output capability. It is also

more efficient and convenient to perform than working on low level PIC code development

or modification of legacy codes of plasma physics modeling.

Large-scale PSII pattern writing has been successfully performed using the large-

scale PSII as a mask-based surface layer property conversion technique, as an alternative to

current deposition-based and ink-based direct write technologies. High dose implantation

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 Conclusions and Future Work 221

was found to produce thin-film-like structure in the substrate surface, which enables the

optical characterization of implantation dose distribution. Compared to AES or SIMS depth

profiling, which require high vacuum environment and are limited to small area analysis,

ellipsometry is fast, convenient, and non-destructive, which is highly desirable for large 3-D

dose distribution studies. It is also suitable for 3-D dose characterization whenever cutting

the target into small AES or SIMS samples is not an option. Small 2 pm features were

also implanted and capped with nitrogen doped SiC >2 layer. TEM study shows that PSII

direct write can be performed to faithfully write pm feature sizes. Smaller features may be

possible if mask and experimental conditions are carefully designed. To our knowledge,

this is the first comprehensive study along these lines.

The PSII system can also be operated in a novel PSII/PVD mode, which produces

large-area, high quality nitrogen-doped SiC >2 films on a variety of substrates with good ad­

hesion. SiC>2 films are important optical coating and insulating layer. It is also a hard

protecting and passivating material. Hydrogen-free and low temperature deposited Si02

are strongly recommended for inter-insulating layers in ULSIs and for gate insulators in

thin-film transistors [23]. Oxynitride multilayers have been found to enhance the ULSI

device performance and reliability [22]. Doping of the Si02 layers with other elements

can produce different layer dielectric constants. The PSII/PVD process offers the ability

of coating sensitive or low-melting point substrates at low temperatures. So far, we have

used nitrogen gas for doping, which is safe to handle, easy to pump, and benign to the vac­

uum system. This also eliminated the source of hydrogen incorporation into the deposited

film. With the PSII capability of PSII/PVD, in situ surface cleaning and interface modifica-

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 Conclusions and Future Work 222

tion, spatial control of nitrogen incorporation, pre- and post-deposition implantation are all

easy to realize. The deposited films were studied by AES depth profiling to study the dis­

tribution of nitrogen in the film and the interface. Ellipsometry was also used to find that

the deposited Si02 layers have excellent optical quality that is very close to that of fused

quartz.

High dose nitrogen ion implantation was performed on a group of transition metals

that are possible candidates for anti-diffusion layers for microelectronics applications. Ti

and stainless steel were given special attention because they are used to make test elec­

trode for FEL electron injection gun. Implantation depth profiles were studied in detail

and correlated with Profile Code calculations and target temperature estimation. Surface

morphology study using optical microscope showed that the highly polished stainless steel

substrate surface did not show any deterioration of surface finish, the surface of Ti and

Mo were "conditioned” and made smoother. It was found that surface oxide/carbide layer,

and the substrate material and structure play an important role in determining implantation

depth profile. AES study of implanted and un-processed Ti substrates also confirmed that

the RFI was operated in a pure inductive mode at the implantation pressure used. There

was no sputtered material from the dielectric window detected by OES or collected on the

Ti substrates processed for several hours by PSII.

Successful implantation of an FEL test electrode was also performed, yielding a uni­

form coating, as indicated by surface color, without any sign of arcing or other kind of

surface imperfection, which are believed to be the sites of strong field emission.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 Conclusions and Future Work 223

The successful operation of the large-scale RFI PSII and the surface analysis results

of implanted samples are important for the effort of suppressing field emission from the

electrodes for FEL electron injection gun. Previous study indicated that nitrogen ion im­

plantation of 1 cm ID test samples reduced field emission [2]. However, it is important to

have the capability of processing large FEL electrode, so that HV breakdown test can be

performed on full-scale electrodes to investigate the effect of nitrogen ion implantation on

field emission. The non-linear relationship between breakdown field and gap voltage, the

Total Voltage Effect, requires such extreme testing [2]. HV breakdown test in large-scale

would provide confidence in successful operation under higher voltage for FEL electron

injection gun.

The PSII/PVD system also provides other possibilities of processing electrode sur­

face, such as coating the electrode surface with nitrogen-doped SiOo. A large-scale elec­

trode processed by this method has already been tested at TJNAF HV breakdown test fa­

cility, showing reduced field emission. The test results showed that the processed electrode

could be operated with a field gradient of 24 MV/m, with a dark current of merely 160 pA;

compared to the applied field gradient of 3 -6 MV/m and dark current of a fraction of micro

amps for the unprocessed electrode. The conditioning time for the processed electrode was

also much shorter. Another possibility is to operate the PSII facility as a PSII/CVD system,

with PSII pulse heating and possibly another supplementary heating source, depositing

DLC films on electrode surface with or without implanting the DLC with nitrogen, to pro­

duce DLC, nitrogen-doped DLC, or carbon nitride films on electrode surface. DLC has

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8.2 Future Work 224

many superior properties compared to SiC> 2 . DLC and carbon nitride films are promising

candidates for field emission suppressing because of their extreme hardness.

8.2 Future Work

The CWM RFI PECVD was quite successful in producing high quality DLC films on

semiconductor and metal substrates. DLC films deposited at a substrate temperature of

400° C were found to have mechanical and optical properties comparable with that of DLC

deposited around 800°C. DLC deposition at even lower temperature is highly desirable

when sensitive substrates need to be coated. Preliminary study also found that the DLC

films deposited have very smooth surfaces and excellent optical properties. Best of all,

AFM study found that the DLC on un-scratched mirror polished Si and as-received Ti-

6A1-4V were highly oriented. The crystalline size is of nanometer range. Oriented DLC

is highly sought after for its better micro-electronic and optical properties. Recently, the

research group at Argonne National Lab also found nanocrystal diamond film ideal for

MEMS device manufacturing [126]. Further study is desirable to optimize the deposition

conditions and to understand the physical and chemical mechanism of oriented DLC CVD.

The PECVD system is also ready for carbon nitride films deposition with its heated stage

and gas feed system.

The successful operation of PSII/PVD/CVD system makes several things possible:

First, a vigorous search for best recipe to use on FEL electrode surface treatment

to suppress field emission. As mentioned above, nitrogen ion implantation of FEL test

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8.2 Future Work 225

electrode was successful to produce uniform and arc-free surface nitridation. With the

collaboration of TJNAF scientists, a full-scale implanted electrode can be tested, and data

be correlated with surface analysis results and implantation parameters, so that the best

implantation recipe can be obtained.

Second, doped Si02 films deposition on FEL electrodes. HV breakdown test at TJ­

NAF showed that the nitrogen-doped Si02 film on a 6 in. FEL electrode has helped to

greatly suppress field emission. Further study should be carried out to optimize the depo­

sition parameters to enhance this electrode performance. Extending the life, reducing the

conditioning time, and increasing the maximum field will all add great value to the $25M

Laser facility. It will also be an enabling technology for small particle accelerators and

other high-voltage applications.

Third, spatial control of implanted species concentration in the films deposited by

PVD or CVD, to produce novel multilayer structures for surface modification.

Fourth, DLC films deposition in the PSII system, by feeding methane and hydrogen

gas mixture to the chamber and possibly adding active sample stage heating. PSII can also

be used to perform pre- or post-deposition nitrogen ion implantation, heat the substrate,

optimize film/substrate interface properties, and to modify the deposited DLC film. With

its superior properties compared to Si02, DLC could be another promising candidate for

FEL electrode surface treatment to reduce field emission.

Fifth, Si02 films can also be deposited on a larger variety of substrates. In addition

to Si, Ti, stainless steel substrates, doped Si02 films have also been deposited onto copper,

silver, tungsten, carbon tape and glass substrates with good adhesion. Possible applications

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. S. 2 Future Work 226

include optical coatings, insulating layers, and hard protective coatings, which all require

further investigation of deposition conditions, film properties, and interface properties.

Sixth, to enable a remotely actionable, automatic switch from pure PSII mode to

PSII/PVD mode for multi-step, multi-layer surface treatment, an (electro-) mechanical

Faraday shield positioning system needs to be designed and built, to control the RF plasma

source to be operated in an inductive mode, or capacitive/inductive mix mode. This patentable

set of innovations will be of great field utility.

Seventh, PSII can be used to perform low-temperature surface modification of poly­

mers, rubbers, and other low melting point materials. Preliminary results in our laboratory

showed that nitrogen ion implantation can ’’write” three dimensional features in glass and in

transparent polymer sheets. These kinds of features can be used for optical signal process­

ing. The addition of a cooling stage can further decrease the substrate temperature, protect

the sensitive substrate, and control lateral diffusion of written features.

Eighth, preliminary tests of the use of PSII for ’’direct-write” device developments

have been performed and reported in this dissertation. Such use may be of importance in the

production of sensors, electronic devices, and MEMS structures. This option has been in

preparation for the use of the JLAB FEL as a major ’’direct-write” tool in the areas outlined

above. Drilling, cutting, photolithography, RTP, and other applications, which have been

sufficiently elaborated in prior FEL-IAB meetings, may benefit greatly from pre-laser or

post-laser PSII treatments. One aspect, for example, is in the control o f’’stiction” for mov­

ing members in microscopic gear and cantilever applications. Numerous other examples

exist and are well known to JLAB personnel involved in the use of the FEL. Specifically,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8.2 Future Work 227

continuing investigations are highly recommended for study the means of using PSII to

process non-metals, to create compound layers with specific mechanical, optical, and elec­

trical properties suitable for MEMS, sensors, and electrical devices, and to pursue detailed

particle-in-cell (MAGIC) modeling of the effects and capabilities of PSII on microscopic

scales.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References

[1] J. P. A. M. Driessen. Low-Temperature Chemical Vapor Deposition o f Titanium Ni­ tride. PhD thesis, Technische Universiteit Delft, 1999.

[2] T. J. Venhaus. Plasma Source Ion Implantation o f High Voltage Electrodes. PhD thesis, College of William and Mary, 1999.

[3] M. L. Green, D. Brasen, K. W. Evans-Lutterodt, L. C. Feldman, K. Krisch, W. Lennard, H. T. Tang, L. Manchanda, and M. T. Tang. Applied Physics Letters , 65(7):848-850, 1994.

[4] Ronald E. Loehman. Characterization o f Ceramics. Manning Publications Co., 3 Lewis Street, Greenwich, CT 06830, 1993.

[5] S. C. Sharma, C. A. Dark, R. C. Hyer, M. Green, T. D. Black, A. R. Chourasia, D. R. Chopra, and K. K. Mishra. Appl. Phys. Lett., 56(18): 1781-1783,1990.

[6] S. H. Kim, Y. S. Park, J-W. Lee, H. J. Park, H. K. Jang, I. S. Yang, and W. S. Yun. Materials Science and Engineering, A209:68-73,1996.

[7] Kazuhiro Suzuki, Atsuhito Sawabe, Hiroaki Yasuda, and Tadao Inuzuka. Appl. Phys. Lett., 50( 12):728—729,1997.

[8] S. Belkouch, D. Landheer, R. Taylor, K. Rajesh, and G. I. Sproule. Materials Re­ search Society Proceedings, Amorphous and Crystalline Insulating Thin Films-1996, pages 151-156,1997.

[9] Steven K. Brierley and Lan Nguyen. Materials Research Society Proceedings, Amor­ phous and Crystalline Insulating Thin Films-1996., pages 145-150,1997.

228

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 229

[10] J. R. Conrad, J. L. Radtke, R. A. Dodd, Frank J. Worzala, and Ngoc C. Tran. Journal o f Applied Physics, 62(11):4591-4596,1987.

[11] S. M. Malik, K. Sridharan, R. P. Fetherston, A. Chen, and J. R. Conrad. J. Vac. Sci. Technol. B, 12(2):843-849,1994.

[12] Michael A. Lieberman and Allan J. Lichtenberg. Principles o f Plasma Discharges and Materials Processing. John Wiley and Sons, Inc., 1994.

[13] J. N. Matossian. J. Vac. Sci. Technol. B, 12(2):850,1994.

[14] M. Tuszewski, J. T. Scheuer, I. H. Campbell, and B. K. Launch. J. Vac. Sci.Technol., 12(2):973—976, 1994.

[15] B. P. Wood, I. Henins, R. J. Gribble, W. A. Reass, R. J. Faehl, M. A. Nastasi, and D. J. Rej. J. Vac. Sci. Technol. B, 12(2):870,1994.

[16] S. Qin and C. Chan. J. Vac. Sci. Technol. B, 12(2):962,1994.

[17] J. Hopwood. Plasma Source Science and Technology, 1:109-116, 1992.

[18] S. P. Bozeman, D. A. Tucker, B. R. Stoner, J. T. Glass, and W. M. Hooke. Appl. Phys. Lett., 66(26):3579—3581, 1995.

[19] T. Munsat, W. M. Hooke, S. P. Bozeman, and S. Washburn. Appl. Phys. Lett.. 66(17):2180-2182,1995.

[20] D. L. Pappas and J. Hopwood. J. Vac. Sci. Technol. A, 12(4): 1576-1582,1994.

[21 ] William L. Warren. Materials Research Society Symposium Proceedings - Materials Development for Direct Write Technologies, 624,2000.

[22] G. Lucovsky. Materials Research Society Symposium Proceedings, Amorphous and Crystalline Insulating Thin Films-1996, 446:67-78,1997.

[23] Y. Uchida, S. Takei, and M. Matsumura. Materials Research Society Symposium Proceedings, 446:21-26,1997.

[24] B. W. Buchholtz and F. M. Kustas. Tribology Transactions, 39(2):330-337,1996.

[25] Fretting and wear resistance coatings for turbine engine applications. Technical Re­ port Report No. ASM-R74-98-01, Analytical Services and Materials, Inc., 1998.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 230

[26] F. P. Bowden and D. Tabor. Friction and Lubrication. Methuen and CO. LTD, 11 New Fetter Lane, London EC4,1967.

[27] T. Spalvins. Lubrication with sputtered MoS2 films: Principles, operation, limita­ tions. Technical Report NASA Technical Memorandum 105292, Lewis Research Center, 1991.

[28] George L Trigg. Encyclopedia o f Applied Physics. VCH Publishers, Inc., 220 East 23rd Street, Suite 909, New York, NY 10010,1993.

[29] David R. Lide. CRC Handbook o f Chemistry and Physics. CRC Press, Inc., 2000 Corporate Blvd., N. W., Boca Raton, Florida, 33431,1992-1993.

[30] R. F. Davis. Diamond Films and Coatings. Noyes Publications, 1993.

[31] K. Komvopoulos, N. Saka, and N. P. Suh. Journal ofTribology, 109:223-231,1987.

[32] E. Cappelli, F. Pinzari, P. Ascarelli, and G. Righini. Diamond and Related Materials, 5:292-298,1996.

[33] S. Christiansen, M. Albrecht, H. P. Strunk, H. Homberger, P. M. Marquis, and J. Franks. J. Mater. Res., 11(8):1934—1942,1996.

[34] C. V. Deshpandey and R. F. Bunshah. J. Vac. Sci. Technol., A7(3):2294-2302,1989.

[35] M. L. Hartsell and L. S. Plano. J. Mater. Res., 9(4):921-926,1994.

[36] W. Zhu, P.C. Yang, J. T. Glass, and F. Arezzo. J. Mater.Res., 10(6):1455—1460,1995.

[37] W. Zhu, F. R. Silvazlian, B. R. Stoner, and J. T. Glass. J. Mater. Res., 10(2):425-430, 1995.

[38] E. I. Meletis, A. Erdemir, and G. R. Fenske. Surface and Coatings Technology, 73:39-45, 1995.

[39] Y. Liu, A. Erdemir, and E. I. Meletis. Surface and Coatings Technology, 82:48 - 56, 1996.

[40] F. M. Kustas. M. S. Misra, R. Wei, and P. J. Wilbur. Tribology Transactions, 36(1):113— 119,1993.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 231

[41] Soon-Cheon Seo, David C. Ingram, and Hugh H. Richardson. J. Vac. Sci. Technol. A, 13(6):2856—2862,1995.

[42] M. Basu S.N. Kund and, K. K. Chattopadhyay, A. B. Maity, S. Chaudhuri, and A. K. Pal. Vacuum, 48(5):435-441,1997.

[43] F. Davanloo, H. Park, and C. B. Collins. J. Mater. Res., 11(8):2042—2050,1996.

[44] C. P. Munson, R. J. Faehl, I. Henins, M. Nastasi, W. A. Reass, D. J. Rej, J. T. Scheuer, K. C. Walter, and B. P. Wood. Surface and Coatings Technology, 84:528-536,1996.

[45] F. Alonso, J. J. Ugarte, D. Sansom, J. L. Viviente, and J. I. Onate. Surface and Coatings Technology, 83:301 -306,1996.

[46] E. W. Roberts. Proceedings o f 20th Aerospace Mechanisms Symposium, Lewis Re­ search Center, Cleveland, Ohio, pages 103-119,1986.

[47] L. E. Seitzman and I. L. Singer. Light weight alloys for aerospace applications, 3:371-378, 1995.

[48] L. E. Seitzman, R. N. Bolster, and I. L. Singer. Surface and Coatings Technology, 78(1): 10—13,1996.

[49] Purtech Inc., Progress St., East Stroudsburg, PA 18301, USA (1-570424-1669).

[50] J. A. Woollam Co. Inc., 650 J St., Suite 39, Lincoln, NE 68508, USA (1402477- 7501).

[51] Guide to Using WVASE32. J. A. Woollam Company, 650 J. St., Suite 39, Lincoln, NE 68508,1997.

[52] S. Wolf and R.N. Tauber. Silicon Processing for the VLSI Era, Volumel: Process Technology. Lattice Press, Sunset Beach, California, 1986.

[53] O. S. Heavens. Optical Properties o f Thin Solid Films. Dover Publications, 1991.

[54] Quad Group, 1815-TS Lewis, Spokane, WA 99224, USA (1-509-4584558).

[55] ASTM G99-95ael Standard Test Method for Wear Testing with a Pin-on-Disk Ap­ paratus, Copyright 2000. American Society for Testing and Materials, West Con- shohocken, PA.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 232

[56] ASTM G77-98 Standard Test Methodfor Ranking Resistance o f Materials to Sliding Wear Using Block-on-Ring Wear Test, Copyright 2000. American Society for Testing and Materials, West Conshohocken, PA.

[57] V. P. Godbole, K. Jagannadham, and J. Narayan. Appl. Phys. Lett., 67(9): 1322-1324, 1995.

[58] B. R. Stoner, G. H. M. Ma, S. D. Wolter, and J. T. Glass. Physical Review B, 45(19): 11067-84, 1992.

[59] S. D. Wolter, B. R. Stoner, J. T. Glass, P. J. Ellis, D. S. Buhaenko, C. E. Jenkins, and P. Southworth. Appl. Phys. Lett., 62(11): 1215—1217,1993.

[60] F. Arezzo, N. Zacchetti, and W. Zhu. J. Appl. Phys., 75( 10):5375—5381,1994.

[61] J. Mencik, D. Munz, E. Quandt, E. R. Weppelmann, and M. V. Swain. J. Mater. Res.. 12(9):2475-2484, 1997.

[62] M. Filonenko-Borodich. Theory o f Elasticity. Dover Publications, Inc., 180 Varick Street, NY, N. Y. 10014,1965.

[63] Jung-Hun Kim, Ho-Jun Lee, Youn-Taeg Kim, Ki-Woong Whang, and Jung-Hoon Joo. J. Vac. Sci. Technol. A, 15(3):564—567,1997.

[64] Hideo Sugai, Kenji Nakamura, and Keiji Suzuki. Jpn. J. Appl. Phys., 33(4B):2189— 2193.1994.

[65] Yukinobu Hikosaka, Moritaka Nakamura, and Hideo Sugai. Jpn. J. Appl. Phys., 33(4B):2157—2163,1994.

[66] D. F. Beale, A. E. Wendt, and L. J. Mahoney. J. Vac. Sci. Technol. A, 12(5):2775- 2779.1994.

[67] J. Hopwood. Plasms Source Science and Technology, 3:460-464,1994.

[68] Ho-Jun Lee, II-Dong Yang, and Ki-Woong Whang. Plasma Source Science and Technology, 5:383-388,1996.

[69] Xianmin Tang and Dennis M. Manos. Plasma Sources Science and Technology, 8:594-602, 1999.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 233

[70] Xianmin Tang. Low Damage Processing and Process Characterization. PhD thesis, College of William and Mary, 2000.

[71] J. Hopwood, C. R. Guamieri, S. J. Whitehair, and J. J. Cuomo. J. Vac. Sci. Technol. A, 11(1):152-156, 1993.

[72] Sune Svanberg. Atomic and Molecular Spectroscopy. Springer-Verlag, Berlin, 1992.

[73] Peter F. Bemath. Spectra o f Atoms and Molecules. Oxford University Press, New York, 1995.

[74] F. Debal, M. Wautelet, J. Bretagne, J. P. Dauchot, and M. Hecq. Plasma Source Science and Technology, 9:152-160,2000.

[75] Tomoaki Hino, Ichiro Fujita, and Masana Nishikawa. Plasma Source Science and Technology, 5:424-428,1996.

[76] V. Scherentz and G. Henrion. Plasma Source Science and Technology, 9:32-36, 2000 .

[77] R. W. B. Pearse and A. G. Gaydon. The Identification o f Molecular Spectra. Chap­ man and Hall Ltd, 733 Third Avenue, NY, NY 10017,4th edition, 1984.

[78] A. V. Blant, O. H. Hughes, T. S. Cheng, S. V. Novikov, and C. T. Foxon. Plasma Source Science and Technology, 9:12-17,2000.

[79] T. E. Sheridan and A. J. Alport. Ion-matrix sheath around a square bar. J. Vac. Sci. Technol. B, 12(2):897-900,1994.

[80] R. Faehl, B. De Voider, and B. Wood. Application of particle-in-cell simulation to plasma source ion implantation. J. Vac. Sci. Technol. B, 12(2):884—888, 1994.

[81] S. Mandl, J. Brutscher, R. Gunzel, and W. Moller. J. Vac. Sci. Technol. B, 14(4):2701- 2706,1996.

[82] S. M. Malik, D. E. Muller, K. Sridharan, R. P. Fetherston, Ngoc Tran, and J. R. Conrad. Journal o f Applied Physics, 77(3):1015—1019,1995.

[83] M. Shamim, J. T. Scheuer, and J. R. Conrad. Journal o f Applied Physics, 69(5):2904- 2908,1991.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 234

[84] J. T. Scheuer, M. Shamim, and J. R. Conrad. Journal o f Applied Physics, 87(3):1241— 1245,1990.

[85] Shu Qin, Yuanshong Zhou, and Chung Chan. IEEE Transaction on Plasma Science, 27(3):766—771, 1999.

[86] Zhaoming Zeng, Xiubo Tian, Baoyin Tang Dixon Tai-Kun Kwok, and Paul K. Chu. IEEE Transaction on Plasma Science, 27(4): 1203-1209, 1999.

[87] Dezhen Wang, Tengcai Ma, and Xinlu Deng. J. Vac. Sci. Technol. B, 12(2):905—909, 1994.

[88] M. Paulus, L. Stals, U. Rude, and B. Rauschenbach. Journal o f Applied Physics, 85(2):761—766,1999.

[89] Jorg Brutscher, Reinhard Gunzel, and Wolfhard Moller. Plasma Source Science and Technology, 5:54-60, 1996.

[90] J. P . Verboncoeur, A. B. Langdon, and N. T. Gladd. Computer physics Communica­ tions, 87:199-211,1995.

[91] Bruce Goplen, Larry Ludeking, and David Smithe. Magic Users Manual. Mission Research Corporation, 1997.

[92] Bruce Goplen, Larry Ludeking, David Smithe, and Gary Warren. Computer Physics Communications, 87:54-86,1995.

[93] Kenneth H. Church.Materials Research Society Symposim Proceedings - Materials Development for Direct Write Technologies, 624,2000.

[94] Herbert Herman, Sanjay Sampath, Richard Gambino, Ashish Patel, and Andrew Dent. Materials Research Society Symposium Proceedings - Materials Development for Direct Write Technology, 624,2000.

[95] S. Y. Tan, R. J. Gambino, R. Goswami, S. Sampath, and H. Herman. Materials Research Society Symposium Proceedings - Materials Development for Direct Write Technologies, 624,2000.

[96] Melissa Orme, Qingbin Liu, Jun Zhu, and Robert Smith. Materials Research Soci­ ety Symposium Proceedings - Materials Developmentfor Direct Write Technologies, 624,2000.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 235

[97] Duane Dimos, Paul G. Glem, Nelson S. Bell, Terry J. Garino, Pin Yang, and Mark A. Rodriguez. Materials Research Society Symposium Proceedings - Materials Devel­ opment for Direct Write Technologies, 624,2000.

[98] David Longo and Robert Hill. Materials Research Society Symposium Proceedings - Materials Development for Direct Write Technologies, 624,2000.

[99] D. B. Chrisey, A. Pique, J. M. Fitz-Gerald, R. C. Y. Auyeung, H. D. Wu, S. Lakeou, and R. Chung. Materials Research Society Symposium Proceedings - Materials De­ velopment for Direct Write Technologies, 624,2000.

[100] N. Reis and B. Derby. Materials Research Society Symposium Proceedings - Mate­ rials Development for Direct Write Technologies, 624,2000.

[101] A. Chen, J. Firmiss, and J. R. Conrad. J. Vac. Sci. Technol. B, 12(2):918,1994.

[ 102] Profile Code Software Instruction Manual. Implant Sciences Corporation, 107 Audubon Road, No. 5, Wakefield, Massachusetts 01880,3.20 edition.

[103] Sorab K. Ghandhi. VLSI Fabrication Principles, Silicon and Gallium Arsenide. John Wiley and Sons, INC., New York, NY 10158-0012, second edition, 1994.

[104] N. Hatzopoulos, D. I. Siapkas, P. L. F. Hemment, and W. Skorupa. Journal o f Applied Physics, 80:4960-4970, 1996.

[105] Y. Ma, T. Yasuda, and G. Lucovsky. Applied Physics Letters, 64( 17):2226—2228, 1994.

[106] S. V. Hattangady, H. Niimi, and G. Lucovsky. J. Vac. Sci. Technol. A, 14(6):3017- 3023,1996.

[107] V. A. C. Haanappel, H. D. Van Corbach, T. Fransen, and P. J. Gelings. Materials Science and Engineering, A167:179—185,1993.

[108] V. V. Afanas’ev, A. Stesmans, A. G. Revesz, and H. L. Hughes. Journal o f Applied Physics, 82(5):2184-2199,1997.

[109] C. Vautey, M. Burgos, and M. Langlet. Thin Solid Fdms, 347:184—194,1999.

[110] S. D. Kovaleski, R. M. Gilgenbach, L. K. Ang, and Y. Y. Lau. Journal o f Applied Physics, 86:7129-7138,1999.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 236

[111] E. N. Glezer and E. Mazu. Applied Physics Letters, 71:882,1997.

[112] P. Baeri, R. Reitano, andN. Marino. Applied Surface Science, 86:128-133,1995.

[113] Sang Woo Lim, Yukihiro Shimogaki, Yoshiaki Nakano, Kunio Tada, and Hiroshi Komiyama. Applied Physics Letters, 68(6):832—834, 1996.

[114] L. Wu and D. M. Manos. Materials Research Society Symposium Proceedings- Materials Development for Direct Write Technologies, 624,2000.

[115] K. Suzuki, K Konishi, K. Nakamura, and H. Sugai. Plasma Source Science and Technology, 9(2): 199-204,2000.

[116] L. Wu, B. C. Holloway, D. Prasad Beesabathina, C. Kalil, and D. M. Manos. Surface and Coatings Technology, 130:207-217,2000.

[117] R. Gunzel, M. Betzl, I. Alphonsa, B. Ganguly, P. I. John, and S. Mukherjee. Surface and Coatings Technology, 112:307-309, 1999.

[118] C. Blawert and B. L. Mordike. Surface and Coatings Technology, 116-119:352-360, 1999.

[119] Z. M. Zeng, T. Zhang, B. Y. Tang, X. B. Tian, and P. K. Chu. Surface and Coatings Technology, 115:234-238,1999.

[120] W. Wang, J. H. Booske, H. L. Liu, S. S. Gearhart, and J. L. Shohet. Journal o f Materials Research, 13(3):726—730, 1998.

[121] J. P. Blanchard. Journal o f Vacuum Science and Technology, 12(2):910-917,1994.

[122] J. P. Holman. Heat Transfer. McGraw-Hill Book Company, New York, 1986.

[123] Saul Dushman. Scientific Foundation o f Vacuum Technique. John Wiley and Sons, Inc., New York, 1962.

[124] David H. Holkeboer, Donald W. Jones, Frank Pagano, and Donald J. Santeler. Vac­ uum Technology and Space Simulation. American Institute of Physics, 335 East 45th Street, New York, NY 10017-3483,1993.

[125] B. Larisch, U. Brusky, and H. J Spies. Surface and Coatings Technology, 116- 119:205-211,1999.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References 237

[ 126] A. R. Krauss D. M. Gruen A. Jayatissa S. Bhattacharya J. Tucek. A. Sumant T. D. Cor­ rigan, 0. Auciello and R. P. H. Chang. Materials Research Society Symposium Pro­ ceedings - Electron-Emissive Materials and Vacuum Microelectronics , 621, 2000.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Vita

Lingling Wu

Lingling was bom in Pingwu, Sichuan Province, China on Oct. 27, 1970. She went

to Beijing University to study in Electronics Department in Sept 1986 and received her

bachelor’s degree in Electronics and Information System in July 1990. She also received

her first master’s degree from Beijing University, in July 1993, in Surface and Thin Films

Physics. She entered the Applied Science Department of the College of William and Mary

in Virginia in August 1994 and received her second master’s degree in science in 1996.

Lingling will continue her research in the Physics Division of Candescent Technologies

Corporation in San Jose, California, where she will work on a novel field emission based

fiat panel display.

238

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.